From ddcc60a085057518ccef1470d5e1ed0f76a1312b Mon Sep 17 00:00:00 2001 From: Ahmed Ibrahim Date: Mon, 17 Nov 2025 17:40:11 -0800 Subject: [PATCH 01/10] Update defaults to gpt-5.1 (#6652) ## Summary - update documentation, example configs, and automation defaults to reference gpt-5.1 / gpt-5.1-codex - bump the CLI and core configuration defaults, model presets, and error messaging to the new models while keeping the model-family/tool coverage for legacy slugs - refresh tests, fixtures, and TUI snapshots so they expect the upgraded defaults ## Testing - `cargo test -p codex-core config::tests::test_precedence_fixture_with_gpt5_profile` ------ [Codex Task](https://chatgpt.com/codex/tasks/task_i_6916c5b3c2b08321ace04ee38604fc6b) --- .github/codex/home/config.toml | 2 +- .github/workflows/issue-deduplicator.yml | 2 +- .../src/protocol/common.rs | 4 +- codex-rs/app-server/README.md | 4 +- codex-rs/app-server/tests/suite/config.rs | 4 +- .../tests/suite/set_default_model.rs | 2 +- .../tests/suite/v2/thread_resume.rs | 6 +- .../app-server/tests/suite/v2/thread_start.rs | 2 +- codex-rs/cli/src/main.rs | 8 +- codex-rs/core/src/client.rs | 8 +- codex-rs/core/src/client_common.rs | 8 +- codex-rs/core/src/config/edit.rs | 14 +- codex-rs/core/src/config/mod.rs | 30 +- codex-rs/core/src/model_family.rs | 2 +- codex-rs/core/tests/suite/apply_patch_cli.rs | 12 +- codex-rs/core/tests/suite/approvals.rs | 306 ++++++++++++++++-- codex-rs/core/tests/suite/client.rs | 11 +- .../core/tests/suite/compact_resume_fork.rs | 14 +- codex-rs/core/tests/suite/grep_files.rs | 2 +- codex-rs/core/tests/suite/json_result.rs | 4 +- codex-rs/core/tests/suite/prompt_caching.rs | 4 +- codex-rs/core/tests/suite/review.rs | 4 +- .../core/tests/suite/shell_serialization.rs | 12 +- codex-rs/core/tests/suite/tool_parallelism.rs | 4 +- codex-rs/core/tests/suite/tools.rs | 8 +- codex-rs/core/tests/suite/truncation.rs | 12 +- codex-rs/core/tests/suite/undo.rs | 4 +- codex-rs/core/tests/suite/unified_exec.rs | 32 +- codex-rs/core/tests/suite/user_shell_cmd.rs | 4 +- codex-rs/docs/codex_mcp_interface.md | 6 +- codex-rs/exec/tests/suite/output_schema.rs | 2 +- codex-rs/exec/tests/suite/resume.rs | 6 +- ...tatus_snapshot_includes_monthly_limit.snap | 2 +- ...s_snapshot_includes_reasoning_details.snap | 34 +- ...s_snapshot_shows_empty_limits_message.snap | 32 +- ...snapshot_shows_missing_limits_message.snap | 32 +- ...s_snapshot_shows_stale_limits_message.snap | 2 +- ...snapshot_truncates_in_narrow_terminal.snap | 34 +- codex-rs/tui/src/status/tests.rs | 16 +- codex-rs/utils/tokenizer/src/lib.rs | 2 +- docs/config.md | 10 +- docs/example-config.md | 10 +- docs/exec.md | 4 +- 43 files changed, 483 insertions(+), 238 deletions(-) diff --git a/.github/codex/home/config.toml b/.github/codex/home/config.toml index 8e800ca068..7b065212a8 100644 --- a/.github/codex/home/config.toml +++ b/.github/codex/home/config.toml @@ -1,3 +1,3 @@ -model = "gpt-5" +model = "gpt-5.1" # Consider setting [mcp_servers] here! diff --git a/.github/workflows/issue-deduplicator.yml b/.github/workflows/issue-deduplicator.yml index c36857cab0..579b6a3685 100644 --- a/.github/workflows/issue-deduplicator.yml +++ b/.github/workflows/issue-deduplicator.yml @@ -46,7 +46,7 @@ jobs: with: openai-api-key: ${{ secrets.CODEX_OPENAI_API_KEY }} allow-users: "*" - model: gpt-5 + model: gpt-5.1 prompt: | You are an assistant that triages new GitHub issues by identifying potential duplicates. diff --git a/codex-rs/app-server-protocol/src/protocol/common.rs b/codex-rs/app-server-protocol/src/protocol/common.rs index 5791a3b5bd..db9bed1112 100644 --- a/codex-rs/app-server-protocol/src/protocol/common.rs +++ b/codex-rs/app-server-protocol/src/protocol/common.rs @@ -524,7 +524,7 @@ mod tests { let request = ClientRequest::NewConversation { request_id: RequestId::Integer(42), params: v1::NewConversationParams { - model: Some("gpt-5-codex".to_string()), + model: Some("gpt-5.1-codex".to_string()), model_provider: None, profile: None, cwd: None, @@ -542,7 +542,7 @@ mod tests { "method": "newConversation", "id": 42, "params": { - "model": "gpt-5-codex", + "model": "gpt-5.1-codex", "modelProvider": null, "profile": null, "cwd": null, diff --git a/codex-rs/app-server/README.md b/codex-rs/app-server/README.md index 3ec5b4517d..5f9b87458f 100644 --- a/codex-rs/app-server/README.md +++ b/codex-rs/app-server/README.md @@ -74,7 +74,7 @@ Start a fresh thread when you need a new Codex conversation. { "method": "thread/start", "id": 10, "params": { // Optionally set config settings. If not specified, will use the user's // current config settings. - "model": "gpt-5-codex", + "model": "gpt-5.1-codex", "cwd": "/Users/me/project", "approvalPolicy": "never", "sandbox": "workspaceWrite", @@ -155,7 +155,7 @@ You can optionally specify config overrides on the new turn. If specified, these "writableRoots": ["/Users/me/project"], "networkAccess": true }, - "model": "gpt-5-codex", + "model": "gpt-5.1-codex", "effort": "medium", "summary": "concise" } } diff --git a/codex-rs/app-server/tests/suite/config.rs b/codex-rs/app-server/tests/suite/config.rs index 227d30bbe3..281d54927b 100644 --- a/codex-rs/app-server/tests/suite/config.rs +++ b/codex-rs/app-server/tests/suite/config.rs @@ -27,7 +27,7 @@ fn create_config_toml(codex_home: &Path) -> std::io::Result<()> { std::fs::write( config_toml, r#" -model = "gpt-5-codex" +model = "gpt-5.1-codex" approval_policy = "on-request" sandbox_mode = "workspace-write" model_reasoning_summary = "detailed" @@ -87,7 +87,7 @@ async fn get_config_toml_parses_all_fields() -> Result<()> { }), forced_chatgpt_workspace_id: Some("12345678-0000-0000-0000-000000000000".into()), forced_login_method: Some(ForcedLoginMethod::Chatgpt), - model: Some("gpt-5-codex".into()), + model: Some("gpt-5.1-codex".into()), model_reasoning_effort: Some(ReasoningEffort::High), model_reasoning_summary: Some(ReasoningSummary::Detailed), model_verbosity: Some(Verbosity::Medium), diff --git a/codex-rs/app-server/tests/suite/set_default_model.rs b/codex-rs/app-server/tests/suite/set_default_model.rs index 0c2aa22999..f3af141c07 100644 --- a/codex-rs/app-server/tests/suite/set_default_model.rs +++ b/codex-rs/app-server/tests/suite/set_default_model.rs @@ -57,7 +57,7 @@ fn create_config_toml(codex_home: &Path) -> std::io::Result<()> { std::fs::write( config_toml, r#" -model = "gpt-5-codex" +model = "gpt-5.1-codex" model_reasoning_effort = "medium" "#, ) diff --git a/codex-rs/app-server/tests/suite/v2/thread_resume.rs b/codex-rs/app-server/tests/suite/v2/thread_resume.rs index c201187860..bda2d14172 100644 --- a/codex-rs/app-server/tests/suite/v2/thread_resume.rs +++ b/codex-rs/app-server/tests/suite/v2/thread_resume.rs @@ -27,7 +27,7 @@ async fn thread_resume_returns_original_thread() -> Result<()> { // Start a thread. let start_id = mcp .send_thread_start_request(ThreadStartParams { - model: Some("gpt-5-codex".to_string()), + model: Some("gpt-5.1-codex".to_string()), ..Default::default() }) .await?; @@ -68,7 +68,7 @@ async fn thread_resume_prefers_path_over_thread_id() -> Result<()> { let start_id = mcp .send_thread_start_request(ThreadStartParams { - model: Some("gpt-5-codex".to_string()), + model: Some("gpt-5.1-codex".to_string()), ..Default::default() }) .await?; @@ -112,7 +112,7 @@ async fn thread_resume_supports_history_and_overrides() -> Result<()> { // Start a thread. let start_id = mcp .send_thread_start_request(ThreadStartParams { - model: Some("gpt-5-codex".to_string()), + model: Some("gpt-5.1-codex".to_string()), ..Default::default() }) .await?; diff --git a/codex-rs/app-server/tests/suite/v2/thread_start.rs b/codex-rs/app-server/tests/suite/v2/thread_start.rs index ac2420916b..a5e4c0d487 100644 --- a/codex-rs/app-server/tests/suite/v2/thread_start.rs +++ b/codex-rs/app-server/tests/suite/v2/thread_start.rs @@ -29,7 +29,7 @@ async fn thread_start_creates_thread_and_emits_started() -> Result<()> { // Start a v2 thread with an explicit model override. let req_id = mcp .send_thread_start_request(ThreadStartParams { - model: Some("gpt-5".to_string()), + model: Some("gpt-5.1".to_string()), ..Default::default() }) .await?; diff --git a/codex-rs/cli/src/main.rs b/codex-rs/cli/src/main.rs index 75a7cb8e44..6a3b24aa98 100644 --- a/codex-rs/cli/src/main.rs +++ b/codex-rs/cli/src/main.rs @@ -761,9 +761,9 @@ mod tests { #[test] fn resume_model_flag_applies_when_no_root_flags() { - let interactive = finalize_from_args(["codex", "resume", "-m", "gpt-5-test"].as_ref()); + let interactive = finalize_from_args(["codex", "resume", "-m", "gpt-5.1-test"].as_ref()); - assert_eq!(interactive.model.as_deref(), Some("gpt-5-test")); + assert_eq!(interactive.model.as_deref(), Some("gpt-5.1-test")); assert!(interactive.resume_picker); assert!(!interactive.resume_last); assert_eq!(interactive.resume_session_id, None); @@ -808,7 +808,7 @@ mod tests { "--ask-for-approval", "on-request", "-m", - "gpt-5-test", + "gpt-5.1-test", "-p", "my-profile", "-C", @@ -819,7 +819,7 @@ mod tests { .as_ref(), ); - assert_eq!(interactive.model.as_deref(), Some("gpt-5-test")); + assert_eq!(interactive.model.as_deref(), Some("gpt-5.1-test")); assert!(interactive.oss); assert_eq!(interactive.config_profile.as_deref(), Some("my-profile")); assert_matches!( diff --git a/codex-rs/core/src/client.rs b/codex-rs/core/src/client.rs index 98775e3d3a..4614458646 100644 --- a/codex-rs/core/src/client.rs +++ b/codex-rs/core/src/client.rs @@ -1225,7 +1225,7 @@ mod tests { #[tokio::test] async fn error_when_error_event() { - let raw_error = r#"{"type":"response.failed","sequence_number":3,"response":{"id":"resp_689bcf18d7f08194bf3440ba62fe05d803fee0cdac429894","object":"response","created_at":1755041560,"status":"failed","background":false,"error":{"code":"rate_limit_exceeded","message":"Rate limit reached for gpt-5 in organization org-AAA on tokens per min (TPM): Limit 30000, Used 22999, Requested 12528. Please try again in 11.054s. Visit https://platform.openai.com/account/rate-limits to learn more."}, "usage":null,"user":null,"metadata":{}}}"#; + let raw_error = r#"{"type":"response.failed","sequence_number":3,"response":{"id":"resp_689bcf18d7f08194bf3440ba62fe05d803fee0cdac429894","object":"response","created_at":1755041560,"status":"failed","background":false,"error":{"code":"rate_limit_exceeded","message":"Rate limit reached for gpt-5.1 in organization org-AAA on tokens per min (TPM): Limit 30000, Used 22999, Requested 12528. Please try again in 11.054s. Visit https://platform.openai.com/account/rate-limits to learn more."}, "usage":null,"user":null,"metadata":{}}}"#; let sse1 = format!("event: response.failed\ndata: {raw_error}\n\n"); let provider = ModelProviderInfo { @@ -1254,7 +1254,7 @@ mod tests { Err(CodexErr::Stream(msg, delay)) => { assert_eq!( msg, - "Rate limit reached for gpt-5 in organization org-AAA on tokens per min (TPM): Limit 30000, Used 22999, Requested 12528. Please try again in 11.054s. Visit https://platform.openai.com/account/rate-limits to learn more." + "Rate limit reached for gpt-5.1 in organization org-AAA on tokens per min (TPM): Limit 30000, Used 22999, Requested 12528. Please try again in 11.054s. Visit https://platform.openai.com/account/rate-limits to learn more." ); assert_eq!(*delay, Some(Duration::from_secs_f64(11.054))); } @@ -1473,7 +1473,7 @@ mod tests { fn test_try_parse_retry_after() { let err = Error { r#type: None, - message: Some("Rate limit reached for gpt-5 in organization org- on tokens per min (TPM): Limit 1, Used 1, Requested 19304. Please try again in 28ms. Visit https://platform.openai.com/account/rate-limits to learn more.".to_string()), + message: Some("Rate limit reached for gpt-5.1 in organization org- on tokens per min (TPM): Limit 1, Used 1, Requested 19304. Please try again in 28ms. Visit https://platform.openai.com/account/rate-limits to learn more.".to_string()), code: Some("rate_limit_exceeded".to_string()), plan_type: None, resets_at: None @@ -1487,7 +1487,7 @@ mod tests { fn test_try_parse_retry_after_no_delay() { let err = Error { r#type: None, - message: Some("Rate limit reached for gpt-5 in organization on tokens per min (TPM): Limit 30000, Used 6899, Requested 24050. Please try again in 1.898s. Visit https://platform.openai.com/account/rate-limits to learn more.".to_string()), + message: Some("Rate limit reached for gpt-5.1 in organization on tokens per min (TPM): Limit 30000, Used 6899, Requested 24050. Please try again in 1.898s. Visit https://platform.openai.com/account/rate-limits to learn more.".to_string()), code: Some("rate_limit_exceeded".to_string()), plan_type: None, resets_at: None diff --git a/codex-rs/core/src/client_common.rs b/codex-rs/core/src/client_common.rs index 9494ffcdf4..09cc769220 100644 --- a/codex-rs/core/src/client_common.rs +++ b/codex-rs/core/src/client_common.rs @@ -427,7 +427,7 @@ mod tests { expects_apply_patch_instructions: false, }, InstructionsTestCase { - slug: "gpt-5-codex", + slug: "gpt-5.1-codex", expects_apply_patch_instructions: false, }, InstructionsTestCase { @@ -457,7 +457,7 @@ mod tests { let input: Vec = vec![]; let tools: Vec = vec![]; let req = ResponsesApiRequest { - model: "gpt-5", + model: "gpt-5.1", instructions: "i", input: &input, tools: &tools, @@ -498,7 +498,7 @@ mod tests { create_text_param_for_request(None, &Some(schema.clone())).expect("text controls"); let req = ResponsesApiRequest { - model: "gpt-5", + model: "gpt-5.1", instructions: "i", input: &input, tools: &tools, @@ -534,7 +534,7 @@ mod tests { let input: Vec = vec![]; let tools: Vec = vec![]; let req = ResponsesApiRequest { - model: "gpt-5", + model: "gpt-5.1", instructions: "i", input: &input, tools: &tools, diff --git a/codex-rs/core/src/config/edit.rs b/codex-rs/core/src/config/edit.rs index 82be47ac39..2d58ffc604 100644 --- a/codex-rs/core/src/config/edit.rs +++ b/codex-rs/core/src/config/edit.rs @@ -584,7 +584,7 @@ mod tests { codex_home, None, &[ConfigEdit::SetModel { - model: Some("gpt-5-codex".to_string()), + model: Some("gpt-5.1-codex".to_string()), effort: Some(ReasoningEffort::High), }], ) @@ -592,7 +592,7 @@ mod tests { let contents = std::fs::read_to_string(codex_home.join(CONFIG_TOML_FILE)).expect("read config"); - let expected = r#"model = "gpt-5-codex" + let expected = r#"model = "gpt-5.1-codex" model_reasoning_effort = "high" "#; assert_eq!(contents, expected); @@ -722,7 +722,7 @@ model = "o5-preview" std::fs::write( codex_home.join(CONFIG_TOML_FILE), r#"[profiles."team a"] -model = "gpt-5-codex" +model = "gpt-5.1-codex" "#, ) .expect("seed"); @@ -972,14 +972,14 @@ B = \"2\" let codex_home = tmp.path().to_path_buf(); ConfigEditsBuilder::new(&codex_home) - .set_model(Some("gpt-5-codex"), Some(ReasoningEffort::High)) + .set_model(Some("gpt-5.1-codex"), Some(ReasoningEffort::High)) .apply() .await .expect("persist"); let contents = std::fs::read_to_string(codex_home.join(CONFIG_TOML_FILE)).expect("read config"); - let expected = r#"model = "gpt-5-codex" + let expected = r#"model = "gpt-5.1-codex" model_reasoning_effort = "high" "#; assert_eq!(contents, expected); @@ -1001,11 +1001,11 @@ model_reasoning_effort = "low" std::fs::read_to_string(codex_home.join(CONFIG_TOML_FILE)).expect("read config"); assert_eq!(contents, initial_expected); - let updated_expected = r#"model = "gpt-5-codex" + let updated_expected = r#"model = "gpt-5.1-codex" model_reasoning_effort = "high" "#; ConfigEditsBuilder::new(codex_home) - .set_model(Some("gpt-5-codex"), Some(ReasoningEffort::High)) + .set_model(Some("gpt-5.1-codex"), Some(ReasoningEffort::High)) .apply_blocking() .expect("persist update"); contents = std::fs::read_to_string(codex_home.join(CONFIG_TOML_FILE)).expect("read config"); diff --git a/codex-rs/core/src/config/mod.rs b/codex-rs/core/src/config/mod.rs index 1bf180dd36..5b57d4dc01 100644 --- a/codex-rs/core/src/config/mod.rs +++ b/codex-rs/core/src/config/mod.rs @@ -62,11 +62,11 @@ pub mod profile; pub mod types; #[cfg(target_os = "windows")] -pub const OPENAI_DEFAULT_MODEL: &str = "gpt-5"; +pub const OPENAI_DEFAULT_MODEL: &str = "gpt-5.1"; #[cfg(not(target_os = "windows"))] -pub const OPENAI_DEFAULT_MODEL: &str = "gpt-5-codex"; -const OPENAI_DEFAULT_REVIEW_MODEL: &str = "gpt-5-codex"; -pub const GPT_5_CODEX_MEDIUM_MODEL: &str = "gpt-5-codex"; +pub const OPENAI_DEFAULT_MODEL: &str = "gpt-5.1-codex"; +const OPENAI_DEFAULT_REVIEW_MODEL: &str = "gpt-5.1-codex"; +pub const GPT_5_CODEX_MEDIUM_MODEL: &str = "gpt-5.1-codex"; /// Maximum number of bytes of the documentation that will be embedded. Larger /// files are *silently truncated* to this size so we do not take up too much of @@ -81,7 +81,7 @@ pub struct Config { /// Optional override of model selection. pub model: String, - /// Model used specifically for review sessions. Defaults to "gpt-5-codex". + /// Model used specifically for review sessions. Defaults to "gpt-5.1-codex". pub review_model: String, pub model_family: ModelFamily, @@ -2616,7 +2616,7 @@ url = "https://example.com/mcp" let codex_home = TempDir::new()?; ConfigEditsBuilder::new(codex_home.path()) - .set_model(Some("gpt-5-codex"), Some(ReasoningEffort::High)) + .set_model(Some("gpt-5.1-codex"), Some(ReasoningEffort::High)) .apply() .await?; @@ -2624,7 +2624,7 @@ url = "https://example.com/mcp" tokio::fs::read_to_string(codex_home.path().join(CONFIG_TOML_FILE)).await?; let parsed: ConfigToml = toml::from_str(&serialized)?; - assert_eq!(parsed.model.as_deref(), Some("gpt-5-codex")); + assert_eq!(parsed.model.as_deref(), Some("gpt-5.1-codex")); assert_eq!(parsed.model_reasoning_effort, Some(ReasoningEffort::High)); Ok(()) @@ -2638,7 +2638,7 @@ url = "https://example.com/mcp" tokio::fs::write( &config_path, r#" -model = "gpt-5-codex" +model = "gpt-5.1-codex" model_reasoning_effort = "medium" [profiles.dev] @@ -2674,7 +2674,7 @@ model = "gpt-4.1" ConfigEditsBuilder::new(codex_home.path()) .with_profile(Some("dev")) - .set_model(Some("gpt-5-codex"), Some(ReasoningEffort::Medium)) + .set_model(Some("gpt-5.1-codex"), Some(ReasoningEffort::Medium)) .apply() .await?; @@ -2686,7 +2686,7 @@ model = "gpt-4.1" .get("dev") .expect("profile should be created"); - assert_eq!(profile.model.as_deref(), Some("gpt-5-codex")); + assert_eq!(profile.model.as_deref(), Some("gpt-5.1-codex")); assert_eq!( profile.model_reasoning_effort, Some(ReasoningEffort::Medium) @@ -2708,7 +2708,7 @@ model = "gpt-4" model_reasoning_effort = "medium" [profiles.prod] -model = "gpt-5-codex" +model = "gpt-5.1-codex" "#, ) .await?; @@ -2737,7 +2737,7 @@ model = "gpt-5-codex" .profiles .get("prod") .and_then(|profile| profile.model.as_deref()), - Some("gpt-5-codex"), + Some("gpt-5.1-codex"), ); Ok(()) @@ -2852,7 +2852,7 @@ model_provider = "openai" approval_policy = "on-failure" [profiles.gpt5] -model = "gpt-5" +model = "gpt-5.1" model_provider = "openai" approval_policy = "on-failure" model_reasoning_effort = "high" @@ -3168,9 +3168,9 @@ model_verbosity = "high" fixture.codex_home(), )?; let expected_gpt5_profile_config = Config { - model: "gpt-5".to_string(), + model: "gpt-5.1".to_string(), review_model: OPENAI_DEFAULT_REVIEW_MODEL.to_string(), - model_family: find_family_for_model("gpt-5").expect("known model slug"), + model_family: find_family_for_model("gpt-5.1").expect("known model slug"), model_context_window: Some(272_000), model_max_output_tokens: Some(128_000), model_auto_compact_token_limit: Some(244_800), diff --git a/codex-rs/core/src/model_family.rs b/codex-rs/core/src/model_family.rs index b46fae4db7..4a10a6295f 100644 --- a/codex-rs/core/src/model_family.rs +++ b/codex-rs/core/src/model_family.rs @@ -132,7 +132,7 @@ pub fn find_family_for_model(slug: &str) -> Option { model_family!(slug, "gpt-4o", needs_special_apply_patch_instructions: true) } else if slug.starts_with("gpt-3.5") { model_family!(slug, "gpt-3.5", needs_special_apply_patch_instructions: true) - } else if slug.starts_with("test-gpt-5-codex") { + } else if slug.starts_with("test-gpt-5") { model_family!( slug, slug, supports_reasoning_summaries: true, diff --git a/codex-rs/core/tests/suite/apply_patch_cli.rs b/codex-rs/core/tests/suite/apply_patch_cli.rs index db20370370..41b074cc3e 100644 --- a/codex-rs/core/tests/suite/apply_patch_cli.rs +++ b/codex-rs/core/tests/suite/apply_patch_cli.rs @@ -87,8 +87,8 @@ async fn apply_patch_cli_multiple_operations_integration( skip_if_no_network!(Ok(())); let harness = apply_patch_harness_with(|config| { - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is valid"); + config.model = "gpt-5.1".to_string(); + config.model_family = find_family_for_model("gpt-5.1").expect("gpt-5.1 is valid"); }) .await?; @@ -671,8 +671,8 @@ async fn apply_patch_shell_heredoc_with_cd_updates_relative_workdir() -> Result< skip_if_no_network!(Ok(())); let harness = apply_patch_harness_with(|config| { - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is valid"); + config.model = "gpt-5.1".to_string(); + config.model_family = find_family_for_model("gpt-5.1").expect("gpt-5.1 is valid"); }) .await?; @@ -717,8 +717,8 @@ async fn apply_patch_shell_failure_propagates_error_and_skips_diff() -> Result<( skip_if_no_network!(Ok(())); let harness = apply_patch_harness_with(|config| { - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is valid"); + config.model = "gpt-5.1".to_string(); + config.model_family = find_family_for_model("gpt-5.1").expect("gpt-5.1 is valid"); }) .await?; let test = harness.test(); diff --git a/codex-rs/core/tests/suite/approvals.rs b/codex-rs/core/tests/suite/approvals.rs index 8316d4de59..a106d2eae0 100644 --- a/codex-rs/core/tests/suite/approvals.rs +++ b/codex-rs/core/tests/suite/approvals.rs @@ -240,6 +240,10 @@ enum Expectation { target: TargetPath, content: &'static str, }, + FileCreatedNoExitCode { + target: TargetPath, + content: &'static str, + }, PatchApplied { target: TargetPath, content: &'static str, @@ -251,12 +255,18 @@ enum Expectation { NetworkSuccess { body_contains: &'static str, }, + NetworkSuccessNoExitCode { + body_contains: &'static str, + }, NetworkFailure { expect_tag: &'static str, }, CommandSuccess { stdout_contains: &'static str, }, + CommandSuccessNoExitCode { + stdout_contains: &'static str, + }, CommandFailure { output_contains: &'static str, }, @@ -270,8 +280,7 @@ impl Expectation { assert_eq!( result.exit_code, Some(0), - "expected successful exit for {:?}", - path + "expected successful exit for {path:?}" ); assert!( result.stdout.contains(content), @@ -285,6 +294,21 @@ impl Expectation { ); let _ = fs::remove_file(path); } + Expectation::FileCreatedNoExitCode { target, content } => { + let (path, _) = target.resolve_for_patch(test); + assert_eq!(result.exit_code, None, "expected no exit code for {path:?}"); + assert!( + result.stdout.contains(content), + "stdout missing {content:?}: {}", + result.stdout + ); + let file_contents = fs::read_to_string(&path)?; + assert!( + file_contents.contains(content), + "file contents missing {content:?}: {file_contents}" + ); + let _ = fs::remove_file(path); + } Expectation::PatchApplied { target, content } => { let (path, _) = target.resolve_for_patch(test); match result.exit_code { @@ -360,6 +384,23 @@ impl Expectation { result.stdout ); } + Expectation::NetworkSuccessNoExitCode { body_contains } => { + assert_eq!( + result.exit_code, None, + "expected no exit code for successful network call: {}", + result.stdout + ); + assert!( + result.stdout.contains("OK:"), + "stdout missing OK prefix: {}", + result.stdout + ); + assert!( + result.stdout.contains(body_contains), + "stdout missing body text {body_contains:?}: {}", + result.stdout + ); + } Expectation::NetworkFailure { expect_tag } => { assert_ne!( result.exit_code, @@ -391,6 +432,18 @@ impl Expectation { result.stdout ); } + Expectation::CommandSuccessNoExitCode { stdout_contains } => { + assert_eq!( + result.exit_code, None, + "expected no exit code for trusted command: {}", + result.stdout + ); + assert!( + result.stdout.contains(stdout_contains), + "trusted command stdout missing {stdout_contains:?}: {}", + result.stdout + ); + } Expectation::CommandFailure { output_contains } => { assert_ne!( result.exit_code, @@ -588,13 +641,30 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::FileCreated { target: TargetPath::OutsideWorkspace("dfa_on_request.txt"), content: "danger-on-request", }, }, + ScenarioSpec { + name: "danger_full_access_on_request_allows_outside_write_gpt_5_1_no_exit", + approval_policy: OnRequest, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::WriteFile { + target: TargetPath::OutsideWorkspace("dfa_on_request_5_1.txt"), + content: "danger-on-request", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::OutsideWorkspace("dfa_on_request_5_1.txt"), + content: "danger-on-request", + }, + }, ScenarioSpec { name: "danger_full_access_on_request_allows_network", approval_policy: OnRequest, @@ -605,12 +675,28 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::NetworkSuccess { body_contains: "danger-network-ok", }, }, + ScenarioSpec { + name: "danger_full_access_on_request_allows_network_gpt_5_1_no_exit", + approval_policy: OnRequest, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::FetchUrl { + endpoint: "/dfa/network", + response_body: "danger-network-ok", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::NetworkSuccessNoExitCode { + body_contains: "danger-network-ok", + }, + }, ScenarioSpec { name: "trusted_command_unless_trusted_runs_without_prompt", approval_policy: UnlessTrusted, @@ -620,12 +706,27 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::CommandSuccess { stdout_contains: "trusted-unless", }, }, + ScenarioSpec { + name: "trusted_command_unless_trusted_runs_without_prompt_gpt_5_1_no_exit", + approval_policy: UnlessTrusted, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::RunCommand { + command: &["echo", "trusted-unless"], + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::CommandSuccessNoExitCode { + stdout_contains: "trusted-unless", + }, + }, ScenarioSpec { name: "danger_full_access_on_failure_allows_outside_write", approval_policy: OnFailure, @@ -636,13 +737,30 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::FileCreated { target: TargetPath::OutsideWorkspace("dfa_on_failure.txt"), content: "danger-on-failure", }, }, + ScenarioSpec { + name: "danger_full_access_on_failure_allows_outside_write_gpt_5_1_no_exit", + approval_policy: OnFailure, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::WriteFile { + target: TargetPath::OutsideWorkspace("dfa_on_failure_5_1.txt"), + content: "danger-on-failure", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::OutsideWorkspace("dfa_on_failure_5_1.txt"), + content: "danger-on-failure", + }, + }, ScenarioSpec { name: "danger_full_access_unless_trusted_requests_approval", approval_policy: UnlessTrusted, @@ -653,7 +771,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -663,6 +781,26 @@ fn scenarios() -> Vec { content: "danger-unless-trusted", }, }, + ScenarioSpec { + name: "danger_full_access_unless_trusted_requests_approval_gpt_5_1_no_exit", + approval_policy: UnlessTrusted, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::WriteFile { + target: TargetPath::OutsideWorkspace("dfa_unless_trusted_5_1.txt"), + content: "danger-unless-trusted", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::ExecApproval { + decision: ReviewDecision::Approved, + expected_reason: None, + }, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::OutsideWorkspace("dfa_unless_trusted_5_1.txt"), + content: "danger-unless-trusted", + }, + }, ScenarioSpec { name: "danger_full_access_never_allows_outside_write", approval_policy: Never, @@ -673,13 +811,30 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::FileCreated { target: TargetPath::OutsideWorkspace("dfa_never.txt"), content: "danger-never", }, }, + ScenarioSpec { + name: "danger_full_access_never_allows_outside_write_gpt_5_1_no_exit", + approval_policy: Never, + sandbox_policy: SandboxPolicy::DangerFullAccess, + action: ActionKind::WriteFile { + target: TargetPath::OutsideWorkspace("dfa_never_5_1.txt"), + content: "danger-never", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::OutsideWorkspace("dfa_never_5_1.txt"), + content: "danger-never", + }, + }, ScenarioSpec { name: "read_only_on_request_requires_approval", approval_policy: OnRequest, @@ -690,7 +845,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: true, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -700,6 +855,26 @@ fn scenarios() -> Vec { content: "read-only-approval", }, }, + ScenarioSpec { + name: "read_only_on_request_requires_approval_gpt_5_1_no_exit", + approval_policy: OnRequest, + sandbox_policy: SandboxPolicy::ReadOnly, + action: ActionKind::WriteFile { + target: TargetPath::Workspace("ro_on_request_5_1.txt"), + content: "read-only-approval", + }, + with_escalated_permissions: true, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::ExecApproval { + decision: ReviewDecision::Approved, + expected_reason: None, + }, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::Workspace("ro_on_request_5_1.txt"), + content: "read-only-approval", + }, + }, ScenarioSpec { name: "trusted_command_on_request_read_only_runs_without_prompt", approval_policy: OnRequest, @@ -709,12 +884,27 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::CommandSuccess { stdout_contains: "trusted-read-only", }, }, + ScenarioSpec { + name: "trusted_command_on_request_read_only_runs_without_prompt_gpt_5_1_no_exit", + approval_policy: OnRequest, + sandbox_policy: SandboxPolicy::ReadOnly, + action: ActionKind::RunCommand { + command: &["echo", "trusted-read-only"], + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::Auto, + expectation: Expectation::CommandSuccessNoExitCode { + stdout_contains: "trusted-read-only", + }, + }, ScenarioSpec { name: "read_only_on_request_blocks_network", approval_policy: OnRequest, @@ -760,7 +950,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: Some("command failed; retry without sandbox?"), @@ -770,6 +960,27 @@ fn scenarios() -> Vec { content: "read-only-on-failure", }, }, + #[cfg(not(target_os = "linux"))] + ScenarioSpec { + name: "read_only_on_failure_escalates_after_sandbox_error_gpt_5_1_no_exit", + approval_policy: OnFailure, + sandbox_policy: SandboxPolicy::ReadOnly, + action: ActionKind::WriteFile { + target: TargetPath::Workspace("ro_on_failure_5_1.txt"), + content: "read-only-on-failure", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::ExecApproval { + decision: ReviewDecision::Approved, + expected_reason: Some("command failed; retry without sandbox?"), + }, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::Workspace("ro_on_failure_5_1.txt"), + content: "read-only-on-failure", + }, + }, ScenarioSpec { name: "read_only_on_request_network_escalates_when_approved", approval_policy: OnRequest, @@ -780,7 +991,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: true, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -789,6 +1000,25 @@ fn scenarios() -> Vec { body_contains: "read-only-network-ok", }, }, + ScenarioSpec { + name: "read_only_on_request_network_escalates_when_approved_gpt_5_1_no_exit", + approval_policy: OnRequest, + sandbox_policy: SandboxPolicy::ReadOnly, + action: ActionKind::FetchUrl { + endpoint: "/ro/network-approved", + response_body: "read-only-network-ok", + }, + with_escalated_permissions: true, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::ExecApproval { + decision: ReviewDecision::Approved, + expected_reason: None, + }, + expectation: Expectation::NetworkSuccessNoExitCode { + body_contains: "read-only-network-ok", + }, + }, ScenarioSpec { name: "apply_patch_shell_requires_patch_approval", approval_policy: UnlessTrusted, @@ -819,7 +1049,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::Auto, expectation: Expectation::PatchApplied { target: TargetPath::Workspace("apply_patch_function.txt"), @@ -836,7 +1066,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![Feature::ApplyPatchFreeform], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::Auto, expectation: Expectation::PatchApplied { target: TargetPath::OutsideWorkspace("apply_patch_function_danger.txt"), @@ -853,7 +1083,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::PatchApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -873,7 +1103,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::PatchApproval { decision: ReviewDecision::Denied, expected_reason: None, @@ -913,7 +1143,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::PatchApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -933,7 +1163,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: Some("gpt-5-codex"), + model_override: Some("gpt-5.1-codex"), outcome: Outcome::Auto, expectation: Expectation::FileNotCreated { target: TargetPath::OutsideWorkspace("apply_patch_function_never.txt"), @@ -952,7 +1182,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -962,6 +1192,26 @@ fn scenarios() -> Vec { content: "read-only-unless-trusted", }, }, + ScenarioSpec { + name: "read_only_unless_trusted_requires_approval_gpt_5_1_no_exit", + approval_policy: UnlessTrusted, + sandbox_policy: SandboxPolicy::ReadOnly, + action: ActionKind::WriteFile { + target: TargetPath::Workspace("ro_unless_trusted_5_1.txt"), + content: "read-only-unless-trusted", + }, + with_escalated_permissions: false, + features: vec![], + model_override: Some("gpt-5.1"), + outcome: Outcome::ExecApproval { + decision: ReviewDecision::Approved, + expected_reason: None, + }, + expectation: Expectation::FileCreatedNoExitCode { + target: TargetPath::Workspace("ro_unless_trusted_5_1.txt"), + content: "read-only-unless-trusted", + }, + }, ScenarioSpec { name: "read_only_never_reports_sandbox_failure", approval_policy: Never, @@ -992,7 +1242,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::CommandSuccess { stdout_contains: "trusted-never", @@ -1008,7 +1258,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::FileCreated { target: TargetPath::Workspace("ww_on_request.txt"), @@ -1039,7 +1289,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: true, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -1059,7 +1309,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::NetworkSuccess { body_contains: "workspace-network-ok", @@ -1076,7 +1326,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: Some("command failed; retry without sandbox?"), @@ -1096,7 +1346,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: None, @@ -1137,7 +1387,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: false, features: vec![Feature::UnifiedExec], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::Auto, expectation: Expectation::CommandSuccess { stdout_contains: "hello unified exec", @@ -1155,7 +1405,7 @@ fn scenarios() -> Vec { }, with_escalated_permissions: true, features: vec![Feature::UnifiedExec], - model_override: None, + model_override: Some("gpt-5"), outcome: Outcome::ExecApproval { decision: ReviewDecision::Approved, expected_reason: Some(DEFAULT_UNIFIED_EXEC_JUSTIFICATION), @@ -1208,7 +1458,7 @@ async fn run_scenario(scenario: &ScenarioSpec) -> Result<()> { let mut builder = test_codex().with_config(move |config| { config.approval_policy = approval_policy; config.sandbox_policy = sandbox_policy.clone(); - let model = model_override.unwrap_or("gpt-5"); + let model = model_override.unwrap_or("gpt-5.1"); config.model = model.to_string(); config.model_family = find_family_for_model(model).expect("model should map to a known family"); diff --git a/codex-rs/core/tests/suite/client.rs b/codex-rs/core/tests/suite/client.rs index 79c158e084..fefb18f327 100644 --- a/codex-rs/core/tests/suite/client.rs +++ b/codex-rs/core/tests/suite/client.rs @@ -769,7 +769,7 @@ async fn configured_verbosity_not_sent_for_models_without_support() -> anyhow::R let resp_mock = responses::mount_sse_once(&server, sse_completed("resp1")).await; let TestCodex { codex, .. } = test_codex() - .with_model("gpt-5-codex") + .with_model("gpt-5.1-codex") .with_config(|config| { config.model_verbosity = Some(Verbosity::High); }) @@ -807,7 +807,7 @@ async fn configured_verbosity_is_sent() -> anyhow::Result<()> { let resp_mock = responses::mount_sse_once(&server, sse_completed("resp1")).await; let TestCodex { codex, .. } = test_codex() - .with_model("gpt-5") + .with_model("gpt-5.1") .with_config(|config| { config.model_verbosity = Some(Verbosity::High); }) @@ -1155,7 +1155,7 @@ async fn token_count_includes_rate_limits_snapshot() { "reasoning_output_tokens": 0, "total_tokens": 123 }, - // Default model is gpt-5-codex in tests → 95% usable context window + // Default model is gpt-5.1-codex in tests → 95% usable context window "model_context_window": 258400 }, "rate_limits": { @@ -1304,8 +1304,9 @@ async fn context_window_error_sets_total_tokens_to_model_window() -> anyhow::Res let TestCodex { codex, .. } = test_codex() .with_config(|config| { - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("known gpt-5 model family"); + config.model = "gpt-5.1".to_string(); + config.model_family = + find_family_for_model("gpt-5.1").expect("known gpt-5.1 model family"); config.model_context_window = Some(272_000); }) .build(&server) diff --git a/codex-rs/core/tests/suite/compact_resume_fork.rs b/codex-rs/core/tests/suite/compact_resume_fork.rs index e10f5748fb..f5ed7772dc 100644 --- a/codex-rs/core/tests/suite/compact_resume_fork.rs +++ b/codex-rs/core/tests/suite/compact_resume_fork.rs @@ -18,7 +18,6 @@ use codex_core::NewConversation; use codex_core::built_in_model_providers; use codex_core::compact::SUMMARIZATION_PROMPT; use codex_core::config::Config; -use codex_core::config::OPENAI_DEFAULT_MODEL; use codex_core::protocol::EventMsg; use codex_core::protocol::Op; use codex_core::protocol::WarningEvent; @@ -111,9 +110,10 @@ async fn compact_resume_and_fork_preserve_model_history_view() { // 1. Arrange mocked SSE responses for the initial compact/resume/fork flow. let server = MockServer::start().await; mount_initial_flow(&server).await; - + let expected_model = "gpt-5.1-codex"; // 2. Start a new conversation and drive it through the compact/resume/fork steps. - let (_home, config, manager, base) = start_test_conversation(&server).await; + let (_home, config, manager, base) = + start_test_conversation(&server, Some(expected_model)).await; user_turn(&base, "hello world").await; compact_conversation(&base).await; @@ -189,7 +189,6 @@ async fn compact_resume_and_fork_preserve_model_history_view() { .as_str() .unwrap_or_default() .to_string(); - let expected_model = OPENAI_DEFAULT_MODEL; let summary_after_compact = extract_summary_message(&requests[2], SUMMARY_TEXT); let summary_after_resume = extract_summary_message(&requests[3], SUMMARY_TEXT); let summary_after_fork = extract_summary_message(&requests[4], SUMMARY_TEXT); @@ -558,7 +557,7 @@ async fn compact_resume_after_second_compaction_preserves_history() { mount_second_compact_flow(&server).await; // 2. Drive the conversation through compact -> resume -> fork -> compact -> resume. - let (_home, config, manager, base) = start_test_conversation(&server).await; + let (_home, config, manager, base) = start_test_conversation(&server, None).await; user_turn(&base, "hello world").await; compact_conversation(&base).await; @@ -808,6 +807,7 @@ async fn mount_second_compact_flow(server: &MockServer) { async fn start_test_conversation( server: &MockServer, + model: Option<&str>, ) -> (TempDir, Config, ConversationManager, Arc) { let model_provider = ModelProviderInfo { base_url: Some(format!("{}/v1", server.uri())), @@ -817,7 +817,9 @@ async fn start_test_conversation( let mut config = load_default_config_for_test(&home); config.model_provider = model_provider; config.compact_prompt = Some(SUMMARIZATION_PROMPT.to_string()); - + if let Some(model) = model { + config.model = model.to_string(); + } let manager = ConversationManager::with_auth(CodexAuth::from_api_key("dummy")); let NewConversation { conversation, .. } = manager .new_conversation(config.clone()) diff --git a/codex-rs/core/tests/suite/grep_files.rs b/codex-rs/core/tests/suite/grep_files.rs index 8b9f2a3b34..6e31ddf486 100644 --- a/codex-rs/core/tests/suite/grep_files.rs +++ b/codex-rs/core/tests/suite/grep_files.rs @@ -11,7 +11,7 @@ use std::collections::HashSet; use std::path::Path; use std::process::Command as StdCommand; -const MODEL_WITH_TOOL: &str = "test-gpt-5-codex"; +const MODEL_WITH_TOOL: &str = "test-gpt-5.1-codex"; fn ripgrep_available() -> bool { StdCommand::new("rg") diff --git a/codex-rs/core/tests/suite/json_result.rs b/codex-rs/core/tests/suite/json_result.rs index cdfa28e0ae..1b9949102e 100644 --- a/codex-rs/core/tests/suite/json_result.rs +++ b/codex-rs/core/tests/suite/json_result.rs @@ -31,12 +31,12 @@ const SCHEMA: &str = r#" #[tokio::test(flavor = "multi_thread", worker_threads = 2)] async fn codex_returns_json_result_for_gpt5() -> anyhow::Result<()> { - codex_returns_json_result("gpt-5".to_string()).await + codex_returns_json_result("gpt-5.1".to_string()).await } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] async fn codex_returns_json_result_for_gpt5_codex() -> anyhow::Result<()> { - codex_returns_json_result("gpt-5-codex".to_string()).await + codex_returns_json_result("gpt-5.1-codex".to_string()).await } async fn codex_returns_json_result(model: String) -> anyhow::Result<()> { diff --git a/codex-rs/core/tests/suite/prompt_caching.rs b/codex-rs/core/tests/suite/prompt_caching.rs index 0d0f52b47a..c858ea20fd 100644 --- a/codex-rs/core/tests/suite/prompt_caching.rs +++ b/codex-rs/core/tests/suite/prompt_caching.rs @@ -160,7 +160,7 @@ async fn prompt_tools_are_consistent_across_requests() -> anyhow::Result<()> { // with the OpenAI schema, so we just verify the tool presence here let tools_by_model: HashMap<&'static str, Vec<&'static str>> = HashMap::from([ ( - "gpt-5", + "gpt-5.1", vec![ "shell", "list_mcp_resources", @@ -183,7 +183,7 @@ async fn prompt_tools_are_consistent_across_requests() -> anyhow::Result<()> { ], ), ( - "gpt-5-codex", + "gpt-5.1-codex", vec![ "shell", "list_mcp_resources", diff --git a/codex-rs/core/tests/suite/review.rs b/codex-rs/core/tests/suite/review.rs index c56e529d6b..9c3f812c29 100644 --- a/codex-rs/core/tests/suite/review.rs +++ b/codex-rs/core/tests/suite/review.rs @@ -364,7 +364,7 @@ async fn review_uses_custom_review_model_from_config() { // Choose a review model different from the main model; ensure it is used. let codex = new_conversation_for_server(&server, &codex_home, |cfg| { cfg.model = "gpt-4.1".to_string(); - cfg.review_model = "gpt-5".to_string(); + cfg.review_model = "gpt-5.1".to_string(); }) .await; @@ -394,7 +394,7 @@ async fn review_uses_custom_review_model_from_config() { // Assert the request body model equals the configured review model let request = &server.received_requests().await.unwrap()[0]; let body = request.body_json::().unwrap(); - assert_eq!(body["model"].as_str().unwrap(), "gpt-5"); + assert_eq!(body["model"].as_str().unwrap(), "gpt-5.1"); server.verify().await; } diff --git a/codex-rs/core/tests/suite/shell_serialization.rs b/codex-rs/core/tests/suite/shell_serialization.rs index 44d637b9e5..58d32c71d6 100644 --- a/codex-rs/core/tests/suite/shell_serialization.rs +++ b/codex-rs/core/tests/suite/shell_serialization.rs @@ -376,9 +376,9 @@ async fn shell_output_reserializes_truncated_content() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5 is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1 is a model family"); }); let test = builder.build(&server).await?; @@ -741,9 +741,9 @@ async fn shell_output_is_structured_for_nonzero_exit() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); config.include_apply_patch_tool = true; }); let test = builder.build(&server).await?; @@ -847,9 +847,9 @@ async fn local_shell_call_output_is_structured() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); config.include_apply_patch_tool = true; }); let test = builder.build(&server).await?; diff --git a/codex-rs/core/tests/suite/tool_parallelism.rs b/codex-rs/core/tests/suite/tool_parallelism.rs index 30ea0f7d2c..807bb9d82c 100644 --- a/codex-rs/core/tests/suite/tool_parallelism.rs +++ b/codex-rs/core/tests/suite/tool_parallelism.rs @@ -57,9 +57,9 @@ async fn run_turn_and_measure(test: &TestCodex, prompt: &str) -> anyhow::Result< #[allow(clippy::expect_used)] async fn build_codex_with_test_tool(server: &wiremock::MockServer) -> anyhow::Result { let mut builder = test_codex().with_config(|config| { - config.model = "test-gpt-5-codex".to_string(); + config.model = "test-gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("test-gpt-5-codex").expect("test-gpt-5-codex model family"); + find_family_for_model("test-gpt-5.1-codex").expect("test-gpt-5.1-codex model family"); }); builder.build(server).await } diff --git a/codex-rs/core/tests/suite/tools.rs b/codex-rs/core/tests/suite/tools.rs index a6086dcbfb..549d6e7d57 100644 --- a/codex-rs/core/tests/suite/tools.rs +++ b/codex-rs/core/tests/suite/tools.rs @@ -197,9 +197,9 @@ async fn sandbox_denied_shell_returns_original_output() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex model family"); }); let fixture = builder.build(&server).await?; @@ -425,8 +425,8 @@ async fn shell_timeout_handles_background_grandchild_stdout() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is a valid model"); + config.model = "gpt-5.1".to_string(); + config.model_family = find_family_for_model("gpt-5.1").expect("gpt-5.1 is a valid model"); config.sandbox_policy = SandboxPolicy::DangerFullAccess; }); let test = builder.build(&server).await?; diff --git a/codex-rs/core/tests/suite/truncation.rs b/codex-rs/core/tests/suite/truncation.rs index 37ee4f3990..200d27e887 100644 --- a/codex-rs/core/tests/suite/truncation.rs +++ b/codex-rs/core/tests/suite/truncation.rs @@ -41,9 +41,9 @@ async fn truncate_function_error_trims_respond_to_model() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { // Use the test model that wires function tools like grep_files - config.model = "test-gpt-5-codex".to_string(); + config.model = "test-gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("test-gpt-5-codex").expect("model family for test model"); + find_family_for_model("test-gpt-5.1-codex").expect("model family for test model"); }); let test = builder.build(&server).await?; @@ -105,9 +105,9 @@ async fn tool_call_output_exceeds_limit_truncated_for_model() -> Result<()> { // Use a model that exposes the generic shell tool. let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); }); let fixture = builder.build(&server).await?; @@ -197,9 +197,9 @@ async fn tool_call_output_truncated_only_once() -> Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); }); let fixture = builder.build(&server).await?; let call_id = "shell-single-truncation"; diff --git a/codex-rs/core/tests/suite/undo.rs b/codex-rs/core/tests/suite/undo.rs index 7883d347f9..6dcb8e81e0 100644 --- a/codex-rs/core/tests/suite/undo.rs +++ b/codex-rs/core/tests/suite/undo.rs @@ -30,8 +30,8 @@ use pretty_assertions::assert_eq; async fn undo_harness() -> Result { TestCodexHarness::with_config(|config: &mut Config| { config.include_apply_patch_tool = true; - config.model = "gpt-5".to_string(); - config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is valid"); + config.model = "gpt-5.1".to_string(); + config.model_family = find_family_for_model("gpt-5.1").expect("gpt-5.1 is valid"); config.features.enable(Feature::GhostCommit); }) .await diff --git a/codex-rs/core/tests/suite/unified_exec.rs b/codex-rs/core/tests/suite/unified_exec.rs index 8c682efaf2..970be5277e 100644 --- a/codex-rs/core/tests/suite/unified_exec.rs +++ b/codex-rs/core/tests/suite/unified_exec.rs @@ -159,7 +159,7 @@ async fn unified_exec_emits_exec_command_begin_event() -> Result<()> { let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_model("gpt-5").with_config(|config| { config.use_experimental_unified_exec_tool = true; config.features.enable(Feature::UnifiedExec); }); @@ -236,7 +236,7 @@ async fn unified_exec_respects_workdir_override() -> Result<()> { let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_model("gpt-5").with_config(|config| { config.use_experimental_unified_exec_tool = true; config.features.enable(Feature::UnifiedExec); }); @@ -288,28 +288,22 @@ async fn unified_exec_respects_workdir_override() -> Result<()> { }) .await?; + let begin_event = wait_for_event_match(&codex, |msg| match msg { + EventMsg::ExecCommandBegin(event) if event.call_id == call_id => Some(event.clone()), + _ => None, + }) + .await; + + assert_eq!( + begin_event.cwd, workdir, + "exec_command cwd should reflect the requested workdir override" + ); + wait_for_event(&codex, |event| matches!(event, EventMsg::TaskComplete(_))).await; let requests = server.received_requests().await.expect("recorded requests"); assert!(!requests.is_empty(), "expected at least one POST request"); - let bodies = requests - .iter() - .map(|req| req.body_json::().expect("request json")) - .collect::>(); - - let outputs = collect_tool_outputs(&bodies)?; - let output = outputs - .get(call_id) - .expect("missing exec_command workdir output"); - let output_text = output.output.trim(); - let output_canonical = std::fs::canonicalize(output_text)?; - let expected_canonical = std::fs::canonicalize(&workdir)?; - assert_eq!( - output_canonical, expected_canonical, - "pwd should reflect the requested workdir override" - ); - Ok(()) } diff --git a/codex-rs/core/tests/suite/user_shell_cmd.rs b/codex-rs/core/tests/suite/user_shell_cmd.rs index 95c6269843..0d42c45c1c 100644 --- a/codex-rs/core/tests/suite/user_shell_cmd.rs +++ b/codex-rs/core/tests/suite/user_shell_cmd.rs @@ -270,9 +270,9 @@ async fn user_shell_command_is_truncated_only_once() -> anyhow::Result<()> { let server = start_mock_server().await; let mut builder = test_codex().with_config(|config| { - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5-codex").expect("gpt-5-codex is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); }); let fixture = builder.build(&server).await?; diff --git a/codex-rs/docs/codex_mcp_interface.md b/codex-rs/docs/codex_mcp_interface.md index 9b3a959438..a7236e363e 100644 --- a/codex-rs/docs/codex_mcp_interface.md +++ b/codex-rs/docs/codex_mcp_interface.md @@ -55,7 +55,7 @@ Start a new session with optional overrides: Request `newConversation` params (subset): -- `model`: string model id (e.g. "o3", "gpt-5", "gpt-5-codex") +- `model`: string model id (e.g. "o3", "gpt-5.1", "gpt-5.1-codex") - `profile`: optional named profile - `cwd`: optional working directory - `approvalPolicy`: `untrusted` | `on-request` | `on-failure` | `never` @@ -119,13 +119,13 @@ For the complete request/response shapes and flow examples, see the [“Auth end ## Example: start and send a message ```json -{ "jsonrpc": "2.0", "id": 1, "method": "newConversation", "params": { "model": "gpt-5", "approvalPolicy": "on-request" } } +{ "jsonrpc": "2.0", "id": 1, "method": "newConversation", "params": { "model": "gpt-5.1", "approvalPolicy": "on-request" } } ``` Server responds: ```json -{ "jsonrpc": "2.0", "id": 1, "result": { "conversationId": "c7b0…", "model": "gpt-5", "rolloutPath": "/path/to/rollout.jsonl" } } +{ "jsonrpc": "2.0", "id": 1, "result": { "conversationId": "c7b0…", "model": "gpt-5.1", "rolloutPath": "/path/to/rollout.jsonl" } } ``` Then send input: diff --git a/codex-rs/exec/tests/suite/output_schema.rs b/codex-rs/exec/tests/suite/output_schema.rs index 485f5e3c79..89a23147bd 100644 --- a/codex-rs/exec/tests/suite/output_schema.rs +++ b/codex-rs/exec/tests/suite/output_schema.rs @@ -37,7 +37,7 @@ async fn exec_includes_output_schema_in_request() -> anyhow::Result<()> { .arg("--output-schema") .arg(&schema_path) .arg("-m") - .arg("gpt-5") + .arg("gpt-5.1") .arg("tell me a joke") .assert() .success(); diff --git a/codex-rs/exec/tests/suite/resume.rs b/codex-rs/exec/tests/suite/resume.rs index 24b8cb0bce..1a56edded3 100644 --- a/codex-rs/exec/tests/suite/resume.rs +++ b/codex-rs/exec/tests/suite/resume.rs @@ -196,7 +196,7 @@ fn exec_resume_preserves_cli_configuration_overrides() -> anyhow::Result<()> { .arg("--sandbox") .arg("workspace-write") .arg("--model") - .arg("gpt-5") + .arg("gpt-5.1") .arg("-C") .arg(env!("CARGO_MANIFEST_DIR")) .arg(&prompt) @@ -218,7 +218,7 @@ fn exec_resume_preserves_cli_configuration_overrides() -> anyhow::Result<()> { .arg("--sandbox") .arg("workspace-write") .arg("--model") - .arg("gpt-5-high") + .arg("gpt-5.1-high") .arg("-C") .arg(env!("CARGO_MANIFEST_DIR")) .arg(&prompt2) @@ -231,7 +231,7 @@ fn exec_resume_preserves_cli_configuration_overrides() -> anyhow::Result<()> { let stderr = String::from_utf8(output.stderr)?; assert!( - stderr.contains("model: gpt-5-high"), + stderr.contains("model: gpt-5.1-high"), "stderr missing model override: {stderr}" ); if cfg!(target_os = "windows") { diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_monthly_limit.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_monthly_limit.snap index 66484261b9..7ceee1466e 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_monthly_limit.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_monthly_limit.snap @@ -10,7 +10,7 @@ expression: sanitized │ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ │ information on rate limits and credits │ │ │ -│ Model: gpt-5-codex (reasoning none, summaries auto) │ +│ Model: gpt-5.1-codex (reasoning none, summaries auto) │ │ Directory: [[workspace]] │ │ Approval: on-request │ │ Sandbox: read-only │ diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_reasoning_details.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_reasoning_details.snap index 4dc951db58..88d16f86a7 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_reasoning_details.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_includes_reasoning_details.snap @@ -4,20 +4,20 @@ expression: sanitized --- /status -╭─────────────────────────────────────────────────────────────────────╮ -│ >_ OpenAI Codex (v0.0.0) │ -│ │ -│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ -│ information on rate limits and credits │ -│ │ -│ Model: gpt-5-codex (reasoning high, summaries detailed) │ -│ Directory: [[workspace]] │ -│ Approval: on-request │ -│ Sandbox: workspace-write │ -│ Agents.md: │ -│ │ -│ Token usage: 1.9K total (1K input + 900 output) │ -│ Context window: 100% left (2.25K used / 272K) │ -│ 5h limit: [██████░░░░░░░░░░░░░░] 28% left (resets 03:14) │ -│ Weekly limit: [███████████░░░░░░░░░] 55% left (resets 03:24) │ -╰─────────────────────────────────────────────────────────────────────╯ +╭───────────────────────────────────────────────────────────────────────╮ +│ >_ OpenAI Codex (v0.0.0) │ +│ │ +│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ +│ information on rate limits and credits │ +│ │ +│ Model: gpt-5.1-codex (reasoning high, summaries detailed) │ +│ Directory: [[workspace]] │ +│ Approval: on-request │ +│ Sandbox: workspace-write │ +│ Agents.md: │ +│ │ +│ Token usage: 1.9K total (1K input + 900 output) │ +│ Context window: 100% left (2.25K used / 272K) │ +│ 5h limit: [██████░░░░░░░░░░░░░░] 28% left (resets 03:14) │ +│ Weekly limit: [███████████░░░░░░░░░] 55% left (resets 03:24) │ +╰───────────────────────────────────────────────────────────────────────╯ diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_empty_limits_message.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_empty_limits_message.snap index 17862db238..6b21a7ec2e 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_empty_limits_message.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_empty_limits_message.snap @@ -4,19 +4,19 @@ expression: sanitized --- /status -╭─────────────────────────────────────────────────────────────────╮ -│ >_ OpenAI Codex (v0.0.0) │ -│ │ -│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ -│ information on rate limits and credits │ -│ │ -│ Model: gpt-5-codex (reasoning none, summaries auto) │ -│ Directory: [[workspace]] │ -│ Approval: on-request │ -│ Sandbox: read-only │ -│ Agents.md: │ -│ │ -│ Token usage: 750 total (500 input + 250 output) │ -│ Context window: 100% left (750 used / 272K) │ -│ Limits: data not available yet │ -╰─────────────────────────────────────────────────────────────────╯ +╭───────────────────────────────────────────────────────────────────╮ +│ >_ OpenAI Codex (v0.0.0) │ +│ │ +│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ +│ information on rate limits and credits │ +│ │ +│ Model: gpt-5.1-codex (reasoning none, summaries auto) │ +│ Directory: [[workspace]] │ +│ Approval: on-request │ +│ Sandbox: read-only │ +│ Agents.md: │ +│ │ +│ Token usage: 750 total (500 input + 250 output) │ +│ Context window: 100% left (750 used / 272K) │ +│ Limits: data not available yet │ +╰───────────────────────────────────────────────────────────────────╯ diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_missing_limits_message.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_missing_limits_message.snap index 17862db238..6b21a7ec2e 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_missing_limits_message.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_missing_limits_message.snap @@ -4,19 +4,19 @@ expression: sanitized --- /status -╭─────────────────────────────────────────────────────────────────╮ -│ >_ OpenAI Codex (v0.0.0) │ -│ │ -│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ -│ information on rate limits and credits │ -│ │ -│ Model: gpt-5-codex (reasoning none, summaries auto) │ -│ Directory: [[workspace]] │ -│ Approval: on-request │ -│ Sandbox: read-only │ -│ Agents.md: │ -│ │ -│ Token usage: 750 total (500 input + 250 output) │ -│ Context window: 100% left (750 used / 272K) │ -│ Limits: data not available yet │ -╰─────────────────────────────────────────────────────────────────╯ +╭───────────────────────────────────────────────────────────────────╮ +│ >_ OpenAI Codex (v0.0.0) │ +│ │ +│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ +│ information on rate limits and credits │ +│ │ +│ Model: gpt-5.1-codex (reasoning none, summaries auto) │ +│ Directory: [[workspace]] │ +│ Approval: on-request │ +│ Sandbox: read-only │ +│ Agents.md: │ +│ │ +│ Token usage: 750 total (500 input + 250 output) │ +│ Context window: 100% left (750 used / 272K) │ +│ Limits: data not available yet │ +╰───────────────────────────────────────────────────────────────────╯ diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_stale_limits_message.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_stale_limits_message.snap index 7d548f4b2e..b566582144 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_stale_limits_message.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_shows_stale_limits_message.snap @@ -10,7 +10,7 @@ expression: sanitized │ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ │ information on rate limits and credits │ │ │ -│ Model: gpt-5-codex (reasoning none, summaries auto) │ +│ Model: gpt-5.1-codex (reasoning none, summaries auto) │ │ Directory: [[workspace]] │ │ Approval: on-request │ │ Sandbox: read-only │ diff --git a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_truncates_in_narrow_terminal.snap b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_truncates_in_narrow_terminal.snap index ea5444d6d8..99dfe1850b 100644 --- a/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_truncates_in_narrow_terminal.snap +++ b/codex-rs/tui/src/status/snapshots/codex_tui__status__tests__status_snapshot_truncates_in_narrow_terminal.snap @@ -4,21 +4,19 @@ expression: sanitized --- /status -╭────────────────────────────────────────────╮ -│ >_ OpenAI Codex (v0.0.0) │ -│ │ -│ Visit https://chatgpt.com/codex/settings/ │ -│ usage for up-to-date │ -│ information on rate limits and credits │ -│ │ -│ Model: gpt-5-codex (reasoning │ -│ Directory: [[workspace]] │ -│ Approval: on-request │ -│ Sandbox: read-only │ -│ Agents.md: │ -│ │ -│ Token usage: 1.9K total (1K input + │ -│ Context window: 100% left (2.25K used / │ -│ 5h limit: [██████░░░░░░░░░░░░░░] │ -│ (resets 03:14) │ -╰────────────────────────────────────────────╯ +╭────────────────────────────────────────────────────────────────────╮ +│ >_ OpenAI Codex (v0.0.0) │ +│ │ +│ Visit https://chatgpt.com/codex/settings/usage for up-to-date │ +│ information on rate limits and credits │ +│ │ +│ Model: gpt-5.1-codex (reasoning high, summaries detail │ +│ Directory: [[workspace]] │ +│ Approval: on-request │ +│ Sandbox: read-only │ +│ Agents.md: │ +│ │ +│ Token usage: 1.9K total (1K input + 900 output) │ +│ Context window: 100% left (2.25K used / 272K) │ +│ 5h limit: [██████░░░░░░░░░░░░░░] 28% left (resets 03:14) │ +╰────────────────────────────────────────────────────────────────────╯ diff --git a/codex-rs/tui/src/status/tests.rs b/codex-rs/tui/src/status/tests.rs index 49a5937288..5120e36f96 100644 --- a/codex-rs/tui/src/status/tests.rs +++ b/codex-rs/tui/src/status/tests.rs @@ -81,7 +81,7 @@ fn reset_at_from(captured_at: &chrono::DateTime, seconds: i64) -> fn status_snapshot_includes_reasoning_details() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_provider_id = "openai".to_string(); config.model_reasoning_effort = Some(ReasoningEffort::High); config.model_reasoning_summary = ReasoningSummary::Detailed; @@ -144,7 +144,7 @@ fn status_snapshot_includes_reasoning_details() { fn status_snapshot_includes_monthly_limit() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_provider_id = "openai".to_string(); config.cwd = PathBuf::from("/workspace/tests"); @@ -194,7 +194,7 @@ fn status_snapshot_includes_monthly_limit() { fn status_card_token_usage_excludes_cached_tokens() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.cwd = PathBuf::from("/workspace/tests"); let auth_manager = test_auth_manager(&config); @@ -232,7 +232,7 @@ fn status_card_token_usage_excludes_cached_tokens() { fn status_snapshot_truncates_in_narrow_terminal() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.model_provider_id = "openai".to_string(); config.model_reasoning_effort = Some(ReasoningEffort::High); config.model_reasoning_summary = ReasoningSummary::Detailed; @@ -270,7 +270,7 @@ fn status_snapshot_truncates_in_narrow_terminal() { Some(&rate_display), captured_at, ); - let mut rendered_lines = render_lines(&composite.display_lines(46)); + let mut rendered_lines = render_lines(&composite.display_lines(70)); if cfg!(windows) { for line in &mut rendered_lines { *line = line.replace('\\', "/"); @@ -285,7 +285,7 @@ fn status_snapshot_truncates_in_narrow_terminal() { fn status_snapshot_shows_missing_limits_message() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.cwd = PathBuf::from("/workspace/tests"); let auth_manager = test_auth_manager(&config); @@ -325,7 +325,7 @@ fn status_snapshot_shows_missing_limits_message() { fn status_snapshot_shows_empty_limits_message() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.cwd = PathBuf::from("/workspace/tests"); let auth_manager = test_auth_manager(&config); @@ -370,7 +370,7 @@ fn status_snapshot_shows_empty_limits_message() { fn status_snapshot_shows_stale_limits_message() { let temp_home = TempDir::new().expect("temp home"); let mut config = test_config(&temp_home); - config.model = "gpt-5-codex".to_string(); + config.model = "gpt-5.1-codex".to_string(); config.cwd = PathBuf::from("/workspace/tests"); let auth_manager = test_auth_manager(&config); diff --git a/codex-rs/utils/tokenizer/src/lib.rs b/codex-rs/utils/tokenizer/src/lib.rs index 1c343e439e..23ab261fa6 100644 --- a/codex-rs/utils/tokenizer/src/lib.rs +++ b/codex-rs/utils/tokenizer/src/lib.rs @@ -158,7 +158,7 @@ mod tests { #[test] fn model_mapping_builds_tokenizer() -> Result<(), TokenizerError> { // Choose a long-standing model alias that maps to cl100k_base. - let tok = Tokenizer::for_model("gpt-5")?; + let tok = Tokenizer::for_model("gpt-5.1")?; let ids = tok.encode("ok", false); let back = tok.decode(&ids)?; assert_eq!(back, "ok"); diff --git a/docs/config.md b/docs/config.md index ed5aed6fc7..51867773a9 100644 --- a/docs/config.md +++ b/docs/config.md @@ -64,7 +64,7 @@ Notes: The model that Codex should use. ```toml -model = "gpt-5" # overrides the default ("gpt-5-codex" on macOS/Linux, "gpt-5" on Windows) +model = "gpt-5.1" # overrides the default ("gpt-5.1-codex" on macOS/Linux, "gpt-5.1" on Windows) ``` ### model_providers @@ -191,7 +191,7 @@ model = "mistral" ### model_reasoning_effort -If the selected model is known to support reasoning (for example: `o3`, `o4-mini`, `codex-*`, `gpt-5`, `gpt-5-codex`), reasoning is enabled by default when using the Responses API. As explained in the [OpenAI Platform documentation](https://platform.openai.com/docs/guides/reasoning?api-mode=responses#get-started-with-reasoning), this can be set to: +If the selected model is known to support reasoning (for example: `o3`, `o4-mini`, `codex-*`, `gpt-5.1`, `gpt-5.1-codex`), reasoning is enabled by default when using the Responses API. As explained in the [OpenAI Platform documentation](https://platform.openai.com/docs/guides/reasoning?api-mode=responses#get-started-with-reasoning), this can be set to: - `"minimal"` - `"low"` @@ -227,7 +227,7 @@ When set, Codex includes a `text` object in the request payload with the configu Example: ```toml -model = "gpt-5" +model = "gpt-5.1" model_verbosity = "low" ``` @@ -818,7 +818,7 @@ Users can specify config values at multiple levels. Order of precedence is as fo 1. custom command-line argument, e.g., `--model o3` 2. as part of a profile, where the `--profile` is specified via a CLI (or in the config file itself) 3. as an entry in `config.toml`, e.g., `model = "o3"` -4. the default value that comes with Codex CLI (i.e., Codex CLI defaults to `gpt-5-codex`) +4. the default value that comes with Codex CLI (i.e., Codex CLI defaults to `gpt-5.1-codex`) ### history @@ -921,7 +921,7 @@ Valid values: | Key | Type / Values | Notes | | ------------------------------------------------ | ----------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------- | -| `model` | string | Model to use (e.g., `gpt-5-codex`). | +| `model` | string | Model to use (e.g., `gpt-5.1-codex`). | | `model_provider` | string | Provider id from `model_providers` (default: `openai`). | | `model_context_window` | number | Context window tokens. | | `model_max_output_tokens` | number | Max output tokens. | diff --git a/docs/example-config.md b/docs/example-config.md index 43a12a3b5e..21b360ccef 100644 --- a/docs/example-config.md +++ b/docs/example-config.md @@ -19,11 +19,11 @@ Use this example configuration as a starting point. For an explanation of each f ################################################################################ # Primary model used by Codex. Default differs by OS; non-Windows defaults here. -# Linux/macOS default: "gpt-5-codex"; Windows default: "gpt-5". -model = "gpt-5-codex" +# Linux/macOS default: "gpt-5.1-codex"; Windows default: "gpt-5.1". +model = "gpt-5.1-codex" -# Model used by the /review feature (code reviews). Default: "gpt-5-codex". -review_model = "gpt-5-codex" +# Model used by the /review feature (code reviews). Default: "gpt-5.1-codex". +review_model = "gpt-5.1-codex" # Provider id selected from [model_providers]. Default: "openai". model_provider = "openai" @@ -315,7 +315,7 @@ mcp_oauth_credentials_store = "auto" [profiles] # [profiles.default] -# model = "gpt-5-codex" +# model = "gpt-5.1-codex" # model_provider = "openai" # approval_policy = "on-request" # sandbox_mode = "read-only" diff --git a/docs/exec.md b/docs/exec.md index 81475555c7..2b5fb10997 100644 --- a/docs/exec.md +++ b/docs/exec.md @@ -99,8 +99,8 @@ codex exec resume --last "Fix use-after-free issues" Only the conversation context is preserved; you must still provide flags to customize Codex behavior. ```shell -codex exec --model gpt-5-codex --json "Review the change, look for use-after-free issues" -codex exec --model gpt-5 --json resume --last "Fix use-after-free issues" +codex exec --model gpt-5.1-codex --json "Review the change, look for use-after-free issues" +codex exec --model gpt-5.1 --json resume --last "Fix use-after-free issues" ``` ## Authentication From 2b7378ac7761756af981a4f896d0a18554258b11 Mon Sep 17 00:00:00 2001 From: Dylan Hurd Date: Mon, 17 Nov 2025 19:10:56 -0800 Subject: [PATCH 02/10] chore(core) Add shell_serialization coverage (#6810) ## Summary Similar to #6545, this PR updates the shell_serialization test suite to cover the various `shell` tool invocations we have. Note that this does not cover unified_exec, which has its own suite of tests. This should provide some test coverage for when we eventually consolidate serialization logic. ## Testing - [x] These are tests --- codex-rs/core/tests/common/test_codex.rs | 9 + codex-rs/core/tests/suite/apply_patch_cli.rs | 4 +- .../core/tests/suite/shell_serialization.rs | 565 +++++++++--------- 3 files changed, 279 insertions(+), 299 deletions(-) diff --git a/codex-rs/core/tests/common/test_codex.rs b/codex-rs/core/tests/common/test_codex.rs index e88c410ece..c213cab5d6 100644 --- a/codex-rs/core/tests/common/test_codex.rs +++ b/codex-rs/core/tests/common/test_codex.rs @@ -38,6 +38,15 @@ pub enum ApplyPatchModelOutput { ShellViaHeredoc, } +/// A collection of different ways the model can output an apply_patch call +#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)] +pub enum ShellModelOutput { + Shell, + ShellCommand, + LocalShell, + // UnifiedExec has its own set of tests +} + pub struct TestCodexBuilder { config_mutators: Vec>, } diff --git a/codex-rs/core/tests/suite/apply_patch_cli.rs b/codex-rs/core/tests/suite/apply_patch_cli.rs index 41b074cc3e..59ecf421e6 100644 --- a/codex-rs/core/tests/suite/apply_patch_cli.rs +++ b/codex-rs/core/tests/suite/apply_patch_cli.rs @@ -29,7 +29,7 @@ use core_test_support::wait_for_event; use serde_json::json; use test_case::test_case; -async fn apply_patch_harness() -> Result { +pub async fn apply_patch_harness() -> Result { apply_patch_harness_with(|_| {}).await } @@ -43,7 +43,7 @@ async fn apply_patch_harness_with( .await } -async fn mount_apply_patch( +pub async fn mount_apply_patch( harness: &TestCodexHarness, call_id: &str, patch: &str, diff --git a/codex-rs/core/tests/suite/shell_serialization.rs b/codex-rs/core/tests/suite/shell_serialization.rs index 58d32c71d6..3b468ce2b7 100644 --- a/codex-rs/core/tests/suite/shell_serialization.rs +++ b/codex-rs/core/tests/suite/shell_serialization.rs @@ -1,14 +1,13 @@ #![cfg(not(target_os = "windows"))] +#![allow(clippy::expect_used)] use anyhow::Result; use codex_core::features::Feature; use codex_core::model_family::find_family_for_model; use codex_core::protocol::SandboxPolicy; use core_test_support::assert_regex_match; -use core_test_support::responses::ev_apply_patch_function_call; use core_test_support::responses::ev_assistant_message; use core_test_support::responses::ev_completed; -use core_test_support::responses::ev_custom_tool_call; use core_test_support::responses::ev_function_call; use core_test_support::responses::ev_local_shell_call; use core_test_support::responses::ev_response_created; @@ -16,12 +15,18 @@ use core_test_support::responses::mount_sse_sequence; use core_test_support::responses::sse; use core_test_support::responses::start_mock_server; use core_test_support::skip_if_no_network; +use core_test_support::test_codex::ApplyPatchModelOutput; +use core_test_support::test_codex::ShellModelOutput; use core_test_support::test_codex::test_codex; use pretty_assertions::assert_eq; use regex_lite::Regex; use serde_json::Value; use serde_json::json; use std::fs; +use test_case::test_case; + +use crate::suite::apply_patch_cli::apply_patch_harness; +use crate::suite::apply_patch_cli::mount_apply_patch; const FIXTURE_JSON: &str = r#"{ "description": "This is an example JSON file.", @@ -35,34 +40,88 @@ const FIXTURE_JSON: &str = r#"{ } "#; +fn shell_responses( + call_id: &str, + command: Vec<&str>, + output_type: ShellModelOutput, +) -> Result> { + match output_type { + ShellModelOutput::ShellCommand => { + let command = shlex::try_join(command)?; + let parameters = json!({ + "command": command, + "timeout_ms": 2_000, + }); + Ok(vec![ + sse(vec![ + ev_response_created("resp-1"), + ev_function_call( + call_id, + "shell_command", + &serde_json::to_string(¶meters)?, + ), + ev_completed("resp-1"), + ]), + sse(vec![ + ev_assistant_message("msg-1", "done"), + ev_completed("resp-2"), + ]), + ]) + } + ShellModelOutput::Shell => { + let parameters = json!({ + "command": command, + "timeout_ms": 2_000, + }); + Ok(vec![ + sse(vec![ + ev_response_created("resp-1"), + ev_function_call(call_id, "shell", &serde_json::to_string(¶meters)?), + ev_completed("resp-1"), + ]), + sse(vec![ + ev_assistant_message("msg-1", "done"), + ev_completed("resp-2"), + ]), + ]) + } + ShellModelOutput::LocalShell => Ok(vec![ + sse(vec![ + ev_response_created("resp-1"), + ev_local_shell_call(call_id, "completed", command), + ev_completed("resp-1"), + ]), + sse(vec![ + ev_assistant_message("msg-1", "done"), + ev_completed("resp-2"), + ]), + ]), + } +} + #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_stays_json_without_freeform_apply_patch() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_stays_json_without_freeform_apply_patch( + output_type: ShellModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.features.disable(Feature::ApplyPatchFreeform); config.model = "gpt-5".to_string(); config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is a model family"); + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; let call_id = "shell-json"; - let args = json!({ - "command": ["/bin/echo", "shell json"], - "timeout_ms": 1_000, - }); - let responses = vec![ - sse(vec![ - ev_response_created("resp-1"), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses(call_id, vec!["/bin/echo", "shell json"], output_type)?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -80,7 +139,6 @@ async fn shell_output_stays_json_without_freeform_apply_patch() -> Result<()> { let mut parsed: Value = serde_json::from_str(output)?; if let Some(metadata) = parsed.get_mut("metadata").and_then(Value::as_object_mut) { - // duration_seconds is non-deterministic; remove it for deep equality let _ = metadata.remove("duration_seconds"); } @@ -102,31 +160,26 @@ async fn shell_output_stays_json_without_freeform_apply_patch() -> Result<()> { } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_is_structured_with_freeform_apply_patch() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_is_structured_with_freeform_apply_patch( + output_type: ShellModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.features.enable(Feature::ApplyPatchFreeform); + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; let call_id = "shell-structured"; - let args = json!({ - "command": ["/bin/echo", "freeform shell"], - "timeout_ms": 1_000, - }); - let responses = vec![ - sse(vec![ - ev_response_created("resp-1"), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses(call_id, vec!["/bin/echo", "freeform shell"], output_type)?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -159,14 +212,23 @@ freeform shell } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_preserves_fixture_json_without_serialization() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_preserves_fixture_json_without_serialization( + output_type: ShellModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.features.disable(Feature::ApplyPatchFreeform); config.model = "gpt-5".to_string(); config.model_family = find_family_for_model("gpt-5").expect("gpt-5 is a model family"); + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; @@ -175,21 +237,11 @@ async fn shell_output_preserves_fixture_json_without_serialization() -> Result<( let fixture_path_str = fixture_path.to_string_lossy().to_string(); let call_id = "shell-json-fixture"; - let args = json!({ - "command": ["/usr/bin/sed", "-n", "p", fixture_path_str], - "timeout_ms": 1_000, - }); - let responses = vec![ - sse(vec![ - ev_response_created("resp-1"), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses( + call_id, + vec!["/usr/bin/sed", "-n", "p", fixture_path_str.as_str()], + output_type, + )?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -232,12 +284,21 @@ async fn shell_output_preserves_fixture_json_without_serialization() -> Result<( } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_structures_fixture_with_serialization() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_structures_fixture_with_serialization( + output_type: ShellModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.features.enable(Feature::ApplyPatchFreeform); + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; @@ -246,21 +307,11 @@ async fn shell_output_structures_fixture_with_serialization() -> Result<()> { let fixture_path_str = fixture_path.to_string_lossy().to_string(); let call_id = "shell-structured-fixture"; - let args = json!({ - "command": ["/usr/bin/sed", "-n", "p", fixture_path_str], - "timeout_ms": 1_000, - }); - let responses = vec![ - sse(vec![ - ev_response_created("resp-1"), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses( + call_id, + vec!["/usr/bin/sed", "-n", "p", fixture_path_str.as_str()], + output_type, + )?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -298,40 +349,26 @@ async fn shell_output_structures_fixture_with_serialization() -> Result<()> { } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_for_freeform_tool_records_duration() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_for_freeform_tool_records_duration( + output_type: ShellModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.include_apply_patch_tool = true; + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; - #[cfg(target_os = "linux")] - let sleep_cmd = vec!["/bin/bash", "-c", "sleep 1"]; - - #[cfg(target_os = "macos")] - let sleep_cmd = vec!["/bin/bash", "-c", "sleep 1"]; - - #[cfg(windows)] - let sleep_cmd = "timeout 1"; - let call_id = "shell-structured"; - let args = json!({ - "command": sleep_cmd, - "timeout_ms": 2_000, - }); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses(call_id, vec!["/bin/bash", "-c", "sleep 1"], output_type)?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -371,33 +408,26 @@ $"#; } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_reserializes_truncated_content() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_reserializes_truncated_content(output_type: ShellModelOutput) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.model = "gpt-5.1-codex".to_string(); config.model_family = - find_family_for_model("gpt-5.1-codex").expect("gpt-5.1 is a model family"); + find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } + let _ = output_type; }); let test = builder.build(&server).await?; let call_id = "shell-truncated"; - let args = json!({ - "command": ["/bin/sh", "-c", "seq 1 400"], - "timeout_ms": 5_000, - }); - let responses = vec![ - sse(vec![ - ev_response_created("resp-1"), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses(call_id, vec!["/bin/sh", "-c", "seq 1 400"], output_type)?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -445,14 +475,16 @@ $"#; } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn apply_patch_custom_tool_output_is_structured() -> Result<()> { +#[test_case(ApplyPatchModelOutput::Freeform)] +#[test_case(ApplyPatchModelOutput::Function)] +#[test_case(ApplyPatchModelOutput::Shell)] +#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)] +async fn apply_patch_custom_tool_output_is_structured( + output_type: ApplyPatchModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); - let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { - config.include_apply_patch_tool = true; - }); - let test = builder.build(&server).await?; + let harness = apply_patch_harness().await?; let call_id = "apply-patch-structured"; let file_name = "structured.txt"; @@ -463,33 +495,17 @@ async fn apply_patch_custom_tool_output_is_structured() -> Result<()> { *** End Patch "# ); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_custom_tool_call(call_id, "apply_patch", &patch), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "done"), - ev_completed("resp-2"), - ]), - ]; - let mock = mount_sse_sequence(&server, responses).await; + mount_apply_patch(&harness, call_id, &patch, "done", output_type).await; - test.submit_turn_with_policy( - "apply the patch via custom tool", - SandboxPolicy::DangerFullAccess, - ) - .await?; + harness + .test() + .submit_turn_with_policy( + "apply the patch via custom tool", + SandboxPolicy::DangerFullAccess, + ) + .await?; - let req = mock - .last_request() - .expect("apply_patch output request recorded"); - let output_item = req.custom_tool_call_output(call_id); - let output = output_item - .get("output") - .and_then(Value::as_str) - .expect("apply_patch output string"); + let output = harness.apply_patch_output(call_id, output_type).await; let expected_pattern = format!( r"(?s)^Exit code: 0 @@ -499,53 +515,39 @@ Success. Updated the following files: A {file_name} ?$" ); - assert_regex_match(&expected_pattern, output); + assert_regex_match(&expected_pattern, output.as_str()); Ok(()) } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn apply_patch_custom_tool_call_creates_file() -> Result<()> { +#[test_case(ApplyPatchModelOutput::Freeform)] +#[test_case(ApplyPatchModelOutput::Function)] +#[test_case(ApplyPatchModelOutput::Shell)] +#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)] +async fn apply_patch_custom_tool_call_creates_file( + output_type: ApplyPatchModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); - let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { - config.include_apply_patch_tool = true; - }); - let test = builder.build(&server).await?; + let harness = apply_patch_harness().await?; let call_id = "apply-patch-add-file"; let file_name = "custom_tool_apply_patch.txt"; let patch = format!( "*** Begin Patch\n*** Add File: {file_name}\n+custom tool content\n*** End Patch\n" ); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_custom_tool_call(call_id, "apply_patch", &patch), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "apply_patch done"), - ev_completed("resp-2"), - ]), - ]; - let mock = mount_sse_sequence(&server, responses).await; + mount_apply_patch(&harness, call_id, &patch, "apply_patch done", output_type).await; - test.submit_turn_with_policy( - "apply the patch via custom tool to create a file", - SandboxPolicy::DangerFullAccess, - ) - .await?; + harness + .test() + .submit_turn_with_policy( + "apply the patch via custom tool to create a file", + SandboxPolicy::DangerFullAccess, + ) + .await?; - let req = mock - .last_request() - .expect("apply_patch output request recorded"); - let output_item = req.custom_tool_call_output(call_id); - let output = output_item - .get("output") - .and_then(Value::as_str) - .expect("apply_patch output string"); + let output = harness.apply_patch_output(call_id, output_type).await; let expected_pattern = format!( r"(?s)^Exit code: 0 @@ -555,9 +557,9 @@ Success. Updated the following files: A {file_name} ?$" ); - assert_regex_match(&expected_pattern, output); + assert_regex_match(&expected_pattern, output.as_str()); - let new_file_path = test.cwd.path().join(file_name); + let new_file_path = harness.path(file_name); let created_contents = fs::read_to_string(&new_file_path)?; assert_eq!( created_contents, "custom tool content\n", @@ -568,49 +570,42 @@ A {file_name} } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn apply_patch_custom_tool_call_updates_existing_file() -> Result<()> { +#[test_case(ApplyPatchModelOutput::Freeform)] +#[test_case(ApplyPatchModelOutput::Function)] +#[test_case(ApplyPatchModelOutput::Shell)] +#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)] +async fn apply_patch_custom_tool_call_updates_existing_file( + output_type: ApplyPatchModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); - let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { - config.include_apply_patch_tool = true; - }); - let test = builder.build(&server).await?; + let harness = apply_patch_harness().await?; let call_id = "apply-patch-update-file"; let file_name = "custom_tool_apply_patch_existing.txt"; - let file_path = test.cwd.path().join(file_name); + let file_path = harness.path(file_name); fs::write(&file_path, "before\n")?; let patch = format!( "*** Begin Patch\n*** Update File: {file_name}\n@@\n-before\n+after\n*** End Patch\n" ); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_custom_tool_call(call_id, "apply_patch", &patch), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "apply_patch update done"), - ev_completed("resp-2"), - ]), - ]; - let mock = mount_sse_sequence(&server, responses).await; - - test.submit_turn_with_policy( - "apply the patch via custom tool to update a file", - SandboxPolicy::DangerFullAccess, + mount_apply_patch( + &harness, + call_id, + &patch, + "apply_patch update done", + output_type, ) - .await?; + .await; - let req = mock - .last_request() - .expect("apply_patch output request recorded"); - let output_item = req.custom_tool_call_output(call_id); - let output = output_item - .get("output") - .and_then(Value::as_str) - .expect("apply_patch output string"); + harness + .test() + .submit_turn_with_policy( + "apply the patch via custom tool to update a file", + SandboxPolicy::DangerFullAccess, + ) + .await?; + + let output = harness.apply_patch_output(call_id, output_type).await; let expected_pattern = format!( r"(?s)^Exit code: 0 @@ -620,7 +615,7 @@ Success. Updated the following files: M {file_name} ?$" ); - assert_regex_match(&expected_pattern, output); + assert_regex_match(&expected_pattern, output.as_str()); let updated_contents = fs::read_to_string(file_path)?; assert_eq!(updated_contents, "after\n", "expected updated file content"); @@ -629,99 +624,83 @@ M {file_name} } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn apply_patch_custom_tool_call_reports_failure_output() -> Result<()> { +#[test_case(ApplyPatchModelOutput::Freeform)] +#[test_case(ApplyPatchModelOutput::Function)] +#[test_case(ApplyPatchModelOutput::Shell)] +#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)] +async fn apply_patch_custom_tool_call_reports_failure_output( + output_type: ApplyPatchModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); - let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { - config.include_apply_patch_tool = true; - }); - let test = builder.build(&server).await?; + let harness = apply_patch_harness().await?; let call_id = "apply-patch-failure"; let missing_file = "missing_custom_tool_apply_patch.txt"; let patch = format!( "*** Begin Patch\n*** Update File: {missing_file}\n@@\n-before\n+after\n*** End Patch\n" ); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_custom_tool_call(call_id, "apply_patch", &patch), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "apply_patch failure done"), - ev_completed("resp-2"), - ]), - ]; - let mock = mount_sse_sequence(&server, responses).await; - - test.submit_turn_with_policy( - "attempt a failing apply_patch via custom tool", - SandboxPolicy::DangerFullAccess, + mount_apply_patch( + &harness, + call_id, + &patch, + "apply_patch failure done", + output_type, ) - .await?; + .await; - let req = mock - .last_request() - .expect("apply_patch output request recorded"); - let output_item = req.custom_tool_call_output(call_id); - let output = output_item - .get("output") - .and_then(Value::as_str) - .expect("apply_patch output string"); + harness + .test() + .submit_turn_with_policy( + "attempt a failing apply_patch via custom tool", + SandboxPolicy::DangerFullAccess, + ) + .await?; + + let output = harness.apply_patch_output(call_id, output_type).await; let expected_output = format!( "apply_patch verification failed: Failed to read file to update {}/{missing_file}: No such file or directory (os error 2)", - test.cwd.path().to_string_lossy() + harness.cwd().to_string_lossy() ); - assert_eq!(output, expected_output); + assert_eq!(output, expected_output.as_str()); Ok(()) } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn apply_patch_function_call_output_is_structured() -> Result<()> { +#[test_case(ApplyPatchModelOutput::Freeform)] +#[test_case(ApplyPatchModelOutput::Function)] +#[test_case(ApplyPatchModelOutput::Shell)] +#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)] +async fn apply_patch_function_call_output_is_structured( + output_type: ApplyPatchModelOutput, +) -> Result<()> { skip_if_no_network!(Ok(())); - let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { - config.include_apply_patch_tool = true; - }); - let test = builder.build(&server).await?; + let harness = apply_patch_harness().await?; let call_id = "apply-patch-function"; let file_name = "function_apply_patch.txt"; let patch = format!("*** Begin Patch\n*** Add File: {file_name}\n+via function call\n*** End Patch\n"); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_apply_patch_function_call(call_id, &patch), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "apply_patch function done"), - ev_completed("resp-2"), - ]), - ]; - let mock = mount_sse_sequence(&server, responses).await; - - test.submit_turn_with_policy( - "apply the patch via function-call apply_patch", - SandboxPolicy::DangerFullAccess, + mount_apply_patch( + &harness, + call_id, + &patch, + "apply_patch function done", + output_type, ) - .await?; - - let req = mock - .last_request() - .expect("apply_patch function output request recorded"); - let output_item = req.function_call_output(call_id); - let output = output_item - .get("output") - .and_then(Value::as_str) - .expect("apply_patch output string"); + .await; + harness + .test() + .submit_turn_with_policy( + "apply the patch via function-call apply_patch", + SandboxPolicy::DangerFullAccess, + ) + .await?; + let output = harness.apply_patch_output(call_id, output_type).await; let expected_pattern = format!( r"(?s)^Exit code: 0 Wall time: [0-9]+(?:\.[0-9]+)? seconds @@ -730,40 +709,32 @@ Success. Updated the following files: A {file_name} ?$" ); - assert_regex_match(&expected_pattern, output); + assert_regex_match(&expected_pattern, output.as_str()); Ok(()) } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] -async fn shell_output_is_structured_for_nonzero_exit() -> Result<()> { +#[test_case(ShellModelOutput::Shell)] +#[test_case(ShellModelOutput::ShellCommand)] +#[test_case(ShellModelOutput::LocalShell)] +async fn shell_output_is_structured_for_nonzero_exit(output_type: ShellModelOutput) -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.model = "gpt-5.1-codex".to_string(); config.model_family = find_family_for_model("gpt-5.1-codex").expect("gpt-5.1-codex is a model family"); config.include_apply_patch_tool = true; + if matches!(output_type, ShellModelOutput::ShellCommand) { + config.features.enable(Feature::ShellCommandTool); + } }); let test = builder.build(&server).await?; let call_id = "shell-nonzero-exit"; - let args = json!({ - "command": ["/bin/sh", "-c", "exit 42"], - "timeout_ms": 1_000, - }); - let responses = vec![ - sse(vec![ - json!({"type": "response.created", "response": {"id": "resp-1"}}), - ev_function_call(call_id, "shell", &serde_json::to_string(&args)?), - ev_completed("resp-1"), - ]), - sse(vec![ - ev_assistant_message("msg-1", "shell failure handled"), - ev_completed("resp-2"), - ]), - ]; + let responses = shell_responses(call_id, vec!["/bin/sh", "-c", "exit 42"], output_type)?; let mock = mount_sse_sequence(&server, responses).await; test.submit_turn_with_policy( @@ -793,7 +764,7 @@ async fn shell_command_output_is_structured() -> Result<()> { skip_if_no_network!(Ok(())); let server = start_mock_server().await; - let mut builder = test_codex().with_config(|config| { + let mut builder = test_codex().with_config(move |config| { config.features.enable(Feature::ShellCommandTool); }); let test = builder.build(&server).await?; From 28ebe1c97a0c7cd01388890c66b6e77286c98518 Mon Sep 17 00:00:00 2001 From: Dylan Hurd Date: Mon, 17 Nov 2025 22:23:53 -0800 Subject: [PATCH 03/10] fix(windows) shell_command on windows, minor parsing (#6811) ## Summary Enables shell_command for windows users, and starts adding some basic command parsing here, to at least remove powershell prefixes. We'll follow this up with command parsing but I wanted to land this change separately with some basic UX. **NOTE**: This implementation parses bash and powershell on both platforms. In theory this is possible, since you can use git bash on windows or powershell on linux. In practice, this may not be worth the complexity of supporting, so I don't feel strongly about the current approach vs. platform-specific branching. ## Testing - [x] Added a bunch of tests - [x] Ran on both windows and os x --- codex-rs/core/src/lib.rs | 1 + codex-rs/core/src/model_family.rs | 2 + codex-rs/core/src/parse_command.rs | 49 +++++++++++++ codex-rs/core/src/powershell.rs | 93 ++++++++++++++++++++++++ codex-rs/core/src/shell.rs | 14 ++-- codex-rs/core/src/tools/spec.rs | 18 ++++- codex-rs/core/tests/suite/model_tools.rs | 14 +++- codex-rs/tui/src/exec_command.rs | 4 +- 8 files changed, 183 insertions(+), 12 deletions(-) create mode 100644 codex-rs/core/src/powershell.rs diff --git a/codex-rs/core/src/lib.rs b/codex-rs/core/src/lib.rs index e684781194..85618ffa2a 100644 --- a/codex-rs/core/src/lib.rs +++ b/codex-rs/core/src/lib.rs @@ -13,6 +13,7 @@ mod client; mod client_common; pub mod codex; mod codex_conversation; +pub mod powershell; pub use codex_conversation::CodexConversation; mod codex_delegate; mod command_safety; diff --git a/codex-rs/core/src/model_family.rs b/codex-rs/core/src/model_family.rs index 4a10a6295f..db37119d1a 100644 --- a/codex-rs/core/src/model_family.rs +++ b/codex-rs/core/src/model_family.rs @@ -161,6 +161,7 @@ pub fn find_family_for_model(slug: &str) -> Option { "list_dir".to_string(), "read_file".to_string(), ], + shell_type: if cfg!(windows) { ConfigShellToolType::ShellCommand } else { ConfigShellToolType::Default }, supports_parallel_tool_calls: true, support_verbosity: true, ) @@ -176,6 +177,7 @@ pub fn find_family_for_model(slug: &str) -> Option { reasoning_summary_format: ReasoningSummaryFormat::Experimental, base_instructions: GPT_5_CODEX_INSTRUCTIONS.to_string(), apply_patch_tool_type: Some(ApplyPatchToolType::Freeform), + shell_type: if cfg!(windows) { ConfigShellToolType::ShellCommand } else { ConfigShellToolType::Default }, support_verbosity: false, ) } else if slug.starts_with("gpt-5.1") { diff --git a/codex-rs/core/src/parse_command.rs b/codex-rs/core/src/parse_command.rs index f9819ba83b..f335347042 100644 --- a/codex-rs/core/src/parse_command.rs +++ b/codex-rs/core/src/parse_command.rs @@ -1,6 +1,7 @@ use crate::bash::extract_bash_command; use crate::bash::try_parse_shell; use crate::bash::try_parse_word_only_commands_sequence; +use crate::powershell::extract_powershell_command; use codex_protocol::parse_command::ParsedCommand; use shlex::split as shlex_split; use shlex::try_join as shlex_try_join; @@ -11,6 +12,11 @@ pub fn shlex_join(tokens: &[String]) -> String { .unwrap_or_else(|_| "".to_string()) } +/// Extracts the shell and script from a command, regardless of platform +pub fn extract_shell_command(command: &[String]) -> Option<(&str, &str)> { + extract_bash_command(command).or_else(|| extract_powershell_command(command)) +} + /// DO NOT REVIEW THIS CODE BY HAND /// This parsing code is quite complex and not easy to hand-modify. /// The easiest way to iterate is to add unit tests and have Codex fix the implementation. @@ -877,6 +883,42 @@ mod tests { }], ); } + + #[test] + fn powershell_command_is_stripped() { + assert_parsed( + &vec_str(&["powershell", "-Command", "Get-ChildItem"]), + vec![ParsedCommand::Unknown { + cmd: "Get-ChildItem".to_string(), + }], + ); + } + + #[test] + fn pwsh_with_noprofile_and_c_alias_is_stripped() { + assert_parsed( + &vec_str(&["pwsh", "-NoProfile", "-c", "Write-Host hi"]), + vec![ParsedCommand::Unknown { + cmd: "Write-Host hi".to_string(), + }], + ); + } + + #[test] + fn powershell_with_path_is_stripped() { + let command = if cfg!(windows) { + "C:\\windows\\System32\\WindowsPowerShell\\v1.0\\powershell.exe" + } else { + "/usr/local/bin/powershell.exe" + }; + + assert_parsed( + &vec_str(&[command, "-NoProfile", "-c", "Write-Host hi"]), + vec![ParsedCommand::Unknown { + cmd: "Write-Host hi".to_string(), + }], + ); + } } pub fn parse_command_impl(command: &[String]) -> Vec { @@ -884,6 +926,12 @@ pub fn parse_command_impl(command: &[String]) -> Vec { return commands; } + if let Some((_, script)) = extract_powershell_command(command) { + return vec![ParsedCommand::Unknown { + cmd: script.to_string(), + }]; + } + let normalized = normalize_tokens(command); let parts = if contains_connectors(&normalized) { @@ -1190,6 +1238,7 @@ fn parse_find_query_and_path(tail: &[String]) -> (Option, Option } fn parse_shell_lc_commands(original: &[String]) -> Option> { + // Only handle bash/zsh here; PowerShell is stripped separately without bash parsing. let (_, script) = extract_bash_command(original)?; if let Some(tree) = try_parse_shell(script) diff --git a/codex-rs/core/src/powershell.rs b/codex-rs/core/src/powershell.rs new file mode 100644 index 0000000000..844866262e --- /dev/null +++ b/codex-rs/core/src/powershell.rs @@ -0,0 +1,93 @@ +use std::path::PathBuf; + +use crate::shell::ShellType; +use crate::shell::detect_shell_type; + +const POWERSHELL_FLAGS: &[&str] = &["-nologo", "-noprofile", "-command", "-c"]; + +/// Extract the PowerShell script body from an invocation such as: +/// +/// - ["pwsh", "-NoProfile", "-Command", "Get-ChildItem -Recurse | Select-String foo"] +/// - ["powershell.exe", "-Command", "Write-Host hi"] +/// - ["powershell", "-NoLogo", "-NoProfile", "-Command", "...script..."] +/// +/// Returns (`shell`, `script`) when the first arg is a PowerShell executable and a +/// `-Command` (or `-c`) flag is present followed by a script string. +pub fn extract_powershell_command(command: &[String]) -> Option<(&str, &str)> { + if command.len() < 3 { + return None; + } + + let shell = &command[0]; + if detect_shell_type(&PathBuf::from(shell)) != Some(ShellType::PowerShell) { + return None; + } + + // Find the first occurrence of -Command (accept common short alias -c as well) + let mut i = 1usize; + while i + 1 < command.len() { + let flag = &command[i]; + // Reject unknown flags + if !POWERSHELL_FLAGS.contains(&flag.to_ascii_lowercase().as_str()) { + return None; + } + if flag.eq_ignore_ascii_case("-Command") || flag.eq_ignore_ascii_case("-c") { + let script = &command[i + 1]; + return Some((shell, script.as_str())); + } + i += 1; + } + None +} + +#[cfg(test)] +mod tests { + use super::extract_powershell_command; + + #[test] + fn extracts_basic_powershell_command() { + let cmd = vec![ + "powershell".to_string(), + "-Command".to_string(), + "Write-Host hi".to_string(), + ]; + let (_shell, script) = extract_powershell_command(&cmd).expect("extract"); + assert_eq!(script, "Write-Host hi"); + } + + #[test] + fn extracts_lowercase_flags() { + let cmd = vec![ + "powershell".to_string(), + "-nologo".to_string(), + "-command".to_string(), + "Write-Host hi".to_string(), + ]; + let (_shell, script) = extract_powershell_command(&cmd).expect("extract"); + assert_eq!(script, "Write-Host hi"); + } + + #[test] + fn extracts_full_path_powershell_command() { + let command = if cfg!(windows) { + "C:\\windows\\System32\\WindowsPowerShell\\v1.0\\powershell.exe".to_string() + } else { + "/usr/local/bin/powershell.exe".to_string() + }; + let cmd = vec![command, "-Command".to_string(), "Write-Host hi".to_string()]; + let (_shell, script) = extract_powershell_command(&cmd).expect("extract"); + assert_eq!(script, "Write-Host hi"); + } + + #[test] + fn extracts_with_noprofile_and_alias() { + let cmd = vec![ + "pwsh".to_string(), + "-NoProfile".to_string(), + "-c".to_string(), + "Get-ChildItem | Select-String foo".to_string(), + ]; + let (_shell, script) = extract_powershell_command(&cmd).expect("extract"); + assert_eq!(script, "Get-ChildItem | Select-String foo"); + } +} diff --git a/codex-rs/core/src/shell.rs b/codex-rs/core/src/shell.rs index ed52491dad..7bfec089c5 100644 --- a/codex-rs/core/src/shell.rs +++ b/codex-rs/core/src/shell.rs @@ -61,10 +61,7 @@ impl Shell { ] } Shell::PowerShell(ps) => { - let mut args = vec![ - ps.shell_path.to_string_lossy().to_string(), - "-NoLogo".to_string(), - ]; + let mut args = vec![ps.shell_path.to_string_lossy().to_string()]; if !use_login_shell { args.push("-NoProfile".to_string()); } @@ -192,7 +189,6 @@ pub fn detect_shell_type(shell_path: &PathBuf) -> Option { Some("powershell") => Some(ShellType::PowerShell), _ => { let shell_name = shell_path.file_stem(); - if let Some(shell_name) = shell_name && shell_name != shell_path { @@ -251,6 +247,14 @@ mod detect_shell_type_tests { detect_shell_type(&PathBuf::from("powershell.exe")), Some(ShellType::PowerShell) ); + assert_eq!( + detect_shell_type(&PathBuf::from(if cfg!(windows) { + "C:\\windows\\System32\\WindowsPowerShell\\v1.0\\powershell.exe" + } else { + "/usr/local/bin/pwsh" + })), + Some(ShellType::PowerShell) + ); assert_eq!( detect_shell_type(&PathBuf::from("pwsh.exe")), Some(ShellType::PowerShell) diff --git a/codex-rs/core/src/tools/spec.rs b/codex-rs/core/src/tools/spec.rs index 28e88c3811..88ddbf5517 100644 --- a/codex-rs/core/src/tools/spec.rs +++ b/codex-rs/core/src/tools/spec.rs @@ -1292,7 +1292,11 @@ mod tests { "gpt-5-codex", &Features::with_defaults(), &[ - "shell", + if cfg!(windows) { + "shell_command" + } else { + "shell" + }, "list_mcp_resources", "list_mcp_resource_templates", "read_mcp_resource", @@ -1309,7 +1313,11 @@ mod tests { "gpt-5.1-codex", &Features::with_defaults(), &[ - "shell", + if cfg!(windows) { + "shell_command" + } else { + "shell" + }, "list_mcp_resources", "list_mcp_resource_templates", "read_mcp_resource", @@ -1384,7 +1392,11 @@ mod tests { "gpt-5.1-codex-mini", &Features::with_defaults(), &[ - "shell", + if cfg!(windows) { + "shell_command" + } else { + "shell" + }, "list_mcp_resources", "list_mcp_resource_templates", "read_mcp_resource", diff --git a/codex-rs/core/tests/suite/model_tools.rs b/codex-rs/core/tests/suite/model_tools.rs index 42bc655f57..28593a55b8 100644 --- a/codex-rs/core/tests/suite/model_tools.rs +++ b/codex-rs/core/tests/suite/model_tools.rs @@ -119,7 +119,12 @@ async fn model_selects_expected_tools() { assert_eq!( gpt5_codex_tools, vec![ - "shell".to_string(), + if cfg!(windows) { + "shell_command" + } else { + "shell" + } + .to_string(), "list_mcp_resources".to_string(), "list_mcp_resource_templates".to_string(), "read_mcp_resource".to_string(), @@ -133,7 +138,12 @@ async fn model_selects_expected_tools() { assert_eq!( gpt51_codex_tools, vec![ - "shell".to_string(), + if cfg!(windows) { + "shell_command" + } else { + "shell" + } + .to_string(), "list_mcp_resources".to_string(), "list_mcp_resource_templates".to_string(), "read_mcp_resource".to_string(), diff --git a/codex-rs/tui/src/exec_command.rs b/codex-rs/tui/src/exec_command.rs index 6f2212b094..8ce6c2632e 100644 --- a/codex-rs/tui/src/exec_command.rs +++ b/codex-rs/tui/src/exec_command.rs @@ -1,7 +1,7 @@ use std::path::Path; use std::path::PathBuf; -use codex_core::bash::extract_bash_command; +use codex_core::parse_command::extract_shell_command; use dirs::home_dir; use shlex::try_join; @@ -10,7 +10,7 @@ pub(crate) fn escape_command(command: &[String]) -> String { } pub(crate) fn strip_bash_lc_and_escape(command: &[String]) -> String { - if let Some((_, script)) = extract_bash_command(command) { + if let Some((_, script)) = extract_shell_command(command) { return script.to_string(); } escape_command(command) From f3d4e210d87374bee766fad32d7d4adb8bfaec01 Mon Sep 17 00:00:00 2001 From: Lael Date: Tue, 18 Nov 2025 15:16:58 +0800 Subject: [PATCH 04/10] :bug: fix(rmcp-client): refresh OAuth tokens using expires_at (#6574) ## Summary - persist OAuth credential expiry timestamps and rehydrate `expires_in` - proactively refresh rmcp OAuth tokens when `expires_at` is near, then persist ## Testing - just fmt - just fix -p codex-rmcp-client - cargo test -p codex-rmcp-client Fixes #6572 --- codex-rs/rmcp-client/src/oauth.rs | 137 ++++++++++++++++-- .../rmcp-client/src/perform_oauth_login.rs | 3 + codex-rs/rmcp-client/src/rmcp_client.rs | 13 ++ 3 files changed, 138 insertions(+), 15 deletions(-) diff --git a/codex-rs/rmcp-client/src/oauth.rs b/codex-rs/rmcp-client/src/oauth.rs index bd6833fca4..f8eafaf23e 100644 --- a/codex-rs/rmcp-client/src/oauth.rs +++ b/codex-rs/rmcp-client/src/oauth.rs @@ -50,6 +50,7 @@ use tokio::sync::Mutex; use crate::find_codex_home::find_codex_home; const KEYRING_SERVICE: &str = "Codex MCP Credentials"; +const REFRESH_SKEW_MILLIS: u64 = 30_000; #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)] pub struct StoredOAuthTokens { @@ -57,6 +58,8 @@ pub struct StoredOAuthTokens { pub url: String, pub client_id: String, pub token_response: WrappedOAuthTokenResponse, + #[serde(default)] + pub expires_at: Option, } /// Determine where Codex should store and read MCP credentials. @@ -113,6 +116,22 @@ pub(crate) fn has_oauth_tokens( Ok(load_oauth_tokens(server_name, url, store_mode)?.is_some()) } +fn refresh_expires_in_from_timestamp(tokens: &mut StoredOAuthTokens) { + let Some(expires_at) = tokens.expires_at else { + return; + }; + + match expires_in_from_timestamp(expires_at) { + Some(seconds) => { + let duration = Duration::from_secs(seconds); + tokens.token_response.0.set_expires_in(Some(&duration)); + } + None => { + tokens.token_response.0.set_expires_in(None); + } + } +} + fn load_oauth_tokens_from_keyring_with_fallback_to_file( keyring_store: &K, server_name: &str, @@ -137,8 +156,9 @@ fn load_oauth_tokens_from_keyring( let key = compute_store_key(server_name, url)?; match keyring_store.load(KEYRING_SERVICE, &key) { Ok(Some(serialized)) => { - let tokens: StoredOAuthTokens = serde_json::from_str(&serialized) + let mut tokens: StoredOAuthTokens = serde_json::from_str(&serialized) .context("failed to deserialize OAuth tokens from keyring")?; + refresh_expires_in_from_timestamp(&mut tokens); Ok(Some(tokens)) } Ok(None) => Ok(None), @@ -286,13 +306,24 @@ impl OAuthPersistor { match maybe_credentials { Some(credentials) => { + let mut last_credentials = self.inner.last_credentials.lock().await; + let new_token_response = WrappedOAuthTokenResponse(credentials.clone()); + let same_token = last_credentials + .as_ref() + .map(|prev| prev.token_response == new_token_response) + .unwrap_or(false); + let expires_at = if same_token { + last_credentials.as_ref().and_then(|prev| prev.expires_at) + } else { + compute_expires_at_millis(&credentials) + }; let stored = StoredOAuthTokens { server_name: self.inner.server_name.clone(), url: self.inner.url.clone(), client_id, - token_response: WrappedOAuthTokenResponse(credentials.clone()), + token_response: new_token_response, + expires_at, }; - let mut last_credentials = self.inner.last_credentials.lock().await; if last_credentials.as_ref() != Some(&stored) { save_oauth_tokens(&self.inner.server_name, &stored, self.inner.store_mode)?; *last_credentials = Some(stored); @@ -317,6 +348,30 @@ impl OAuthPersistor { Ok(()) } + + pub(crate) async fn refresh_if_needed(&self) -> Result<()> { + let expires_at = { + let guard = self.inner.last_credentials.lock().await; + guard.as_ref().and_then(|tokens| tokens.expires_at) + }; + + if !token_needs_refresh(expires_at) { + return Ok(()); + } + + { + let manager = self.inner.authorization_manager.clone(); + let guard = manager.lock().await; + guard.refresh_token().await.with_context(|| { + format!( + "failed to refresh OAuth tokens for server {}", + self.inner.server_name + ) + })?; + } + + self.persist_if_needed().await + } } const FALLBACK_FILENAME: &str = ".credentials.json"; @@ -366,19 +421,14 @@ fn load_oauth_tokens_from_file(server_name: &str, url: &str) -> Result Result<()> { let mut store = read_fallback_file()?.unwrap_or_default(); let token_response = &tokens.token_response.0; + let expires_at = tokens + .expires_at + .or_else(|| compute_expires_at_millis(token_response)); let refresh_token = token_response .refresh_token() .map(|token| token.secret().to_string()); @@ -403,7 +456,7 @@ fn save_oauth_tokens_to_file(tokens: &StoredOAuthTokens) -> Result<()> { server_url: tokens.url.clone(), client_id: tokens.client_id.clone(), access_token: token_response.access_token().secret().to_string(), - expires_at: compute_expires_at_millis(token_response), + expires_at, refresh_token, scopes, }; @@ -427,7 +480,7 @@ fn delete_oauth_tokens_from_file(key: &str) -> Result { Ok(removed) } -fn compute_expires_at_millis(response: &OAuthTokenResponse) -> Option { +pub(crate) fn compute_expires_at_millis(response: &OAuthTokenResponse) -> Option { let expires_in = response.expires_in()?; let now = SystemTime::now() .duration_since(UNIX_EPOCH) @@ -454,6 +507,19 @@ fn expires_in_from_timestamp(expires_at: u64) -> Option { } } +fn token_needs_refresh(expires_at: Option) -> bool { + let Some(expires_at) = expires_at else { + return false; + }; + + let now = SystemTime::now() + .duration_since(UNIX_EPOCH) + .unwrap_or_else(|_| Duration::from_secs(0)) + .as_millis() as u64; + + now.saturating_add(REFRESH_SKEW_MILLIS) >= expires_at +} + fn compute_store_key(server_name: &str, server_url: &str) -> Result { let mut payload = JsonMap::new(); payload.insert( @@ -589,8 +655,9 @@ mod tests { store.save(KEYRING_SERVICE, &key, &serialized)?; let loaded = - super::load_oauth_tokens_from_keyring(&store, &tokens.server_name, &tokens.url)?; - assert_eq!(loaded, Some(expected)); + super::load_oauth_tokens_from_keyring(&store, &tokens.server_name, &tokens.url)? + .expect("tokens should load from keyring"); + assert_tokens_match_without_expiry(&loaded, &expected); Ok(()) } @@ -750,6 +817,43 @@ mod tests { Ok(()) } + #[test] + fn refresh_expires_in_from_timestamp_restores_future_durations() { + let mut tokens = sample_tokens(); + let expires_at = tokens.expires_at.expect("expires_at should be set"); + + tokens.token_response.0.set_expires_in(None); + super::refresh_expires_in_from_timestamp(&mut tokens); + + let actual = tokens + .token_response + .0 + .expires_in() + .expect("expires_in should be restored") + .as_secs(); + let expected = super::expires_in_from_timestamp(expires_at) + .expect("expires_at should still be in the future"); + let diff = actual.abs_diff(expected); + assert!(diff <= 1, "expires_in drift too large: diff={diff}"); + } + + #[test] + fn refresh_expires_in_from_timestamp_clears_expired_tokens() { + let mut tokens = sample_tokens(); + let now = SystemTime::now() + .duration_since(UNIX_EPOCH) + .unwrap_or_else(|_| Duration::from_secs(0)); + let expired_at = now.as_millis() as u64; + tokens.expires_at = Some(expired_at.saturating_sub(1000)); + + let duration = Duration::from_secs(600); + tokens.token_response.0.set_expires_in(Some(&duration)); + + super::refresh_expires_in_from_timestamp(&mut tokens); + + assert!(tokens.token_response.0.expires_in().is_none()); + } + fn assert_tokens_match_without_expiry( actual: &StoredOAuthTokens, expected: &StoredOAuthTokens, @@ -757,6 +861,7 @@ mod tests { assert_eq!(actual.server_name, expected.server_name); assert_eq!(actual.url, expected.url); assert_eq!(actual.client_id, expected.client_id); + assert_eq!(actual.expires_at, expected.expires_at); assert_token_response_match_without_expiry( &actual.token_response, &expected.token_response, @@ -803,12 +908,14 @@ mod tests { ])); let expires_in = Duration::from_secs(3600); response.set_expires_in(Some(&expires_in)); + let expires_at = super::compute_expires_at_millis(&response); StoredOAuthTokens { server_name: "test-server".to_string(), url: "https://example.test".to_string(), client_id: "client-id".to_string(), token_response: WrappedOAuthTokenResponse(response), + expires_at, } } } diff --git a/codex-rs/rmcp-client/src/perform_oauth_login.rs b/codex-rs/rmcp-client/src/perform_oauth_login.rs index 425e124d7d..d8ffdd3949 100644 --- a/codex-rs/rmcp-client/src/perform_oauth_login.rs +++ b/codex-rs/rmcp-client/src/perform_oauth_login.rs @@ -17,6 +17,7 @@ use urlencoding::decode; use crate::OAuthCredentialsStoreMode; use crate::StoredOAuthTokens; use crate::WrappedOAuthTokenResponse; +use crate::oauth::compute_expires_at_millis; use crate::save_oauth_tokens; use crate::utils::apply_default_headers; use crate::utils::build_default_headers; @@ -91,11 +92,13 @@ pub async fn perform_oauth_login( let credentials = credentials_opt.ok_or_else(|| anyhow!("OAuth provider did not return credentials"))?; + let expires_at = compute_expires_at_millis(&credentials); let stored = StoredOAuthTokens { server_name: server_name.to_string(), url: server_url.to_string(), client_id, token_response: WrappedOAuthTokenResponse(credentials), + expires_at, }; save_oauth_tokens(server_name, &stored, store_mode)?; diff --git a/codex-rs/rmcp-client/src/rmcp_client.rs b/codex-rs/rmcp-client/src/rmcp_client.rs index f21859dc92..d7d3477b00 100644 --- a/codex-rs/rmcp-client/src/rmcp_client.rs +++ b/codex-rs/rmcp-client/src/rmcp_client.rs @@ -267,6 +267,7 @@ impl RmcpClient { params: Option, timeout: Option, ) -> Result { + self.refresh_oauth_if_needed().await; let service = self.service().await?; let rmcp_params = params .map(convert_to_rmcp::<_, PaginatedRequestParam>) @@ -284,6 +285,7 @@ impl RmcpClient { params: Option, timeout: Option, ) -> Result { + self.refresh_oauth_if_needed().await; let service = self.service().await?; let rmcp_params = params .map(convert_to_rmcp::<_, PaginatedRequestParam>) @@ -301,6 +303,7 @@ impl RmcpClient { params: Option, timeout: Option, ) -> Result { + self.refresh_oauth_if_needed().await; let service = self.service().await?; let rmcp_params = params .map(convert_to_rmcp::<_, PaginatedRequestParam>) @@ -318,6 +321,7 @@ impl RmcpClient { params: ReadResourceRequestParams, timeout: Option, ) -> Result { + self.refresh_oauth_if_needed().await; let service = self.service().await?; let rmcp_params: ReadResourceRequestParam = convert_to_rmcp(params)?; let fut = service.read_resource(rmcp_params); @@ -333,6 +337,7 @@ impl RmcpClient { arguments: Option, timeout: Option, ) -> Result { + self.refresh_oauth_if_needed().await; let service = self.service().await?; let params = CallToolRequestParams { arguments, name }; let rmcp_params: CallToolRequestParam = convert_to_rmcp(params)?; @@ -371,6 +376,14 @@ impl RmcpClient { warn!("failed to persist OAuth tokens: {error}"); } } + + async fn refresh_oauth_if_needed(&self) { + if let Some(runtime) = self.oauth_persistor().await + && let Err(error) = runtime.refresh_if_needed().await + { + warn!("failed to refresh OAuth tokens: {error}"); + } + } } async fn create_oauth_transport_and_runtime( From fc55fd7a819189a078e1e7d8f569bd40d745a49e Mon Sep 17 00:00:00 2001 From: jif-oai Date: Tue, 18 Nov 2025 15:26:09 +0000 Subject: [PATCH 05/10] feat: git branch tooling (#6831) --- codex-rs/utils/git/src/branch.rs | 144 +++++++++++++++++++++++++++++++ codex-rs/utils/git/src/lib.rs | 2 + 2 files changed, 146 insertions(+) create mode 100644 codex-rs/utils/git/src/branch.rs diff --git a/codex-rs/utils/git/src/branch.rs b/codex-rs/utils/git/src/branch.rs new file mode 100644 index 0000000000..543dffd9c9 --- /dev/null +++ b/codex-rs/utils/git/src/branch.rs @@ -0,0 +1,144 @@ +use std::ffi::OsString; +use std::path::Path; + +use crate::GitToolingError; +use crate::operations::ensure_git_repository; +use crate::operations::resolve_head; +use crate::operations::resolve_repository_root; +use crate::operations::run_git_for_stdout; + +/// Returns the merge-base commit between `HEAD` and the provided branch, if both exist. +/// +/// The function mirrors `git merge-base HEAD ` but returns `Ok(None)` when +/// the repository has no `HEAD` yet or when the branch cannot be resolved. +pub fn merge_base_with_head( + repo_path: &Path, + branch: &str, +) -> Result, GitToolingError> { + ensure_git_repository(repo_path)?; + let repo_root = resolve_repository_root(repo_path)?; + let head = match resolve_head(repo_root.as_path())? { + Some(head) => head, + None => return Ok(None), + }; + + let branch_ref = match run_git_for_stdout( + repo_root.as_path(), + vec![ + OsString::from("rev-parse"), + OsString::from("--verify"), + OsString::from(branch), + ], + None, + ) { + Ok(rev) => rev, + Err(GitToolingError::GitCommand { .. }) => return Ok(None), + Err(other) => return Err(other), + }; + + let merge_base = run_git_for_stdout( + repo_root.as_path(), + vec![ + OsString::from("merge-base"), + OsString::from(head), + OsString::from(branch_ref), + ], + None, + )?; + + Ok(Some(merge_base)) +} + +#[cfg(test)] +mod tests { + use super::merge_base_with_head; + use crate::GitToolingError; + use pretty_assertions::assert_eq; + use std::path::Path; + use std::process::Command; + use tempfile::tempdir; + + fn run_git_in(repo_path: &Path, args: &[&str]) { + let status = Command::new("git") + .current_dir(repo_path) + .args(args) + .status() + .expect("git command"); + assert!(status.success(), "git command failed: {args:?}"); + } + + fn run_git_stdout(repo_path: &Path, args: &[&str]) -> String { + let output = Command::new("git") + .current_dir(repo_path) + .args(args) + .output() + .expect("git command"); + assert!(output.status.success(), "git command failed: {args:?}"); + String::from_utf8_lossy(&output.stdout).trim().to_string() + } + + fn init_test_repo(repo_path: &Path) { + run_git_in(repo_path, &["init", "--initial-branch=main"]); + run_git_in(repo_path, &["config", "core.autocrlf", "false"]); + } + + fn commit(repo_path: &Path, message: &str) { + run_git_in( + repo_path, + &[ + "-c", + "user.name=Tester", + "-c", + "user.email=test@example.com", + "commit", + "-m", + message, + ], + ); + } + + #[test] + fn merge_base_returns_shared_commit() -> Result<(), GitToolingError> { + let temp = tempdir()?; + let repo = temp.path(); + init_test_repo(repo); + + std::fs::write(repo.join("base.txt"), "base\n")?; + run_git_in(repo, &["add", "base.txt"]); + commit(repo, "base commit"); + + run_git_in(repo, &["checkout", "-b", "feature"]); + std::fs::write(repo.join("feature.txt"), "feature change\n")?; + run_git_in(repo, &["add", "feature.txt"]); + commit(repo, "feature commit"); + + run_git_in(repo, &["checkout", "main"]); + std::fs::write(repo.join("main.txt"), "main change\n")?; + run_git_in(repo, &["add", "main.txt"]); + commit(repo, "main commit"); + + run_git_in(repo, &["checkout", "feature"]); + + let expected = run_git_stdout(repo, &["merge-base", "HEAD", "main"]); + let merge_base = merge_base_with_head(repo, "main")?; + assert_eq!(merge_base, Some(expected)); + + Ok(()) + } + + #[test] + fn merge_base_returns_none_when_branch_missing() -> Result<(), GitToolingError> { + let temp = tempdir()?; + let repo = temp.path(); + init_test_repo(repo); + + std::fs::write(repo.join("tracked.txt"), "tracked\n")?; + run_git_in(repo, &["add", "tracked.txt"]); + commit(repo, "initial"); + + let merge_base = merge_base_with_head(repo, "missing-branch")?; + assert_eq!(merge_base, None); + + Ok(()) + } +} diff --git a/codex-rs/utils/git/src/lib.rs b/codex-rs/utils/git/src/lib.rs index cf2887aa8f..e96b0eab6a 100644 --- a/codex-rs/utils/git/src/lib.rs +++ b/codex-rs/utils/git/src/lib.rs @@ -2,6 +2,7 @@ use std::fmt; use std::path::PathBuf; mod apply; +mod branch; mod errors; mod ghost_commits; mod operations; @@ -13,6 +14,7 @@ pub use apply::apply_git_patch; pub use apply::extract_paths_from_patch; pub use apply::parse_git_apply_output; pub use apply::stage_paths; +pub use branch::merge_base_with_head; pub use errors::GitToolingError; pub use ghost_commits::CreateGhostCommitOptions; pub use ghost_commits::create_ghost_commit; From c20df79a389b016ec718566e38d3c169c0a7cce1 Mon Sep 17 00:00:00 2001 From: jif-oai Date: Tue, 18 Nov 2025 16:05:49 +0000 Subject: [PATCH 06/10] nit: mark ghost commit as stable (#6833) --- codex-rs/core/src/features.rs | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-) diff --git a/codex-rs/core/src/features.rs b/codex-rs/core/src/features.rs index 87788d428d..a161056a0b 100644 --- a/codex-rs/core/src/features.rs +++ b/codex-rs/core/src/features.rs @@ -27,6 +27,8 @@ pub enum Stage { /// Unique features toggled via configuration. #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)] pub enum Feature { + /// Create a ghost commit at each turn. + GhostCommit, /// Use the single unified PTY-backed exec tool. UnifiedExec, /// Use the shell command tool that takes `command` as a single string of @@ -42,8 +44,6 @@ pub enum Feature { WebSearchRequest, /// Enable the model-based risk assessments for sandboxed commands. SandboxCommandAssessment, - /// Create a ghost commit at each turn. - GhostCommit, /// Enable Windows sandbox (restricted token) on Windows. WindowsSandbox, /// Enable the default shell tool. @@ -249,6 +249,14 @@ pub struct FeatureSpec { } pub const FEATURES: &[FeatureSpec] = &[ + // Stable features. + FeatureSpec { + id: Feature::GhostCommit, + key: "undo", + stage: Stage::Stable, + default_enabled: true, + }, + // Unstable features. FeatureSpec { id: Feature::UnifiedExec, key: "unified_exec", @@ -291,12 +299,6 @@ pub const FEATURES: &[FeatureSpec] = &[ stage: Stage::Experimental, default_enabled: false, }, - FeatureSpec { - id: Feature::GhostCommit, - key: "ghost_commit", - stage: Stage::Experimental, - default_enabled: true, - }, FeatureSpec { id: Feature::WindowsSandbox, key: "enable_experimental_windows_sandbox", From 0eb2e6f9ee66f4d67584822f5bec91835ef0d4fe Mon Sep 17 00:00:00 2001 From: jif-oai Date: Tue, 18 Nov 2025 16:34:13 +0000 Subject: [PATCH 07/10] nit: app server (#6830) --- .../app-server/src/codex_message_processor.rs | 14 ++++---- codex-rs/app-server/src/fuzzy_file_search.rs | 4 +++ codex-rs/app-server/src/lib.rs | 2 +- codex-rs/app-server/src/outgoing_message.rs | 32 +++++++++++++------ 4 files changed, 35 insertions(+), 17 deletions(-) diff --git a/codex-rs/app-server/src/codex_message_processor.rs b/codex-rs/app-server/src/codex_message_processor.rs index c5ac6a14af..c5fa2a7fa2 100644 --- a/codex-rs/app-server/src/codex_message_processor.rs +++ b/codex-rs/app-server/src/codex_message_processor.rs @@ -158,8 +158,8 @@ struct ActiveLogin { login_id: Uuid, } -impl ActiveLogin { - fn drop(&self) { +impl Drop for ActiveLogin { + fn drop(&mut self) { self.shutdown_handle.shutdown(); } } @@ -417,7 +417,7 @@ impl CodexMessageProcessor { { let mut guard = self.active_login.lock().await; if let Some(active) = guard.take() { - active.drop(); + drop(active); } } @@ -525,7 +525,7 @@ impl CodexMessageProcessor { { let mut guard = self.active_login.lock().await; if let Some(existing) = guard.take() { - existing.drop(); + drop(existing); } *guard = Some(ActiveLogin { shutdown_handle: shutdown_handle.clone(), @@ -615,7 +615,7 @@ impl CodexMessageProcessor { { let mut guard = self.active_login.lock().await; if let Some(existing) = guard.take() { - existing.drop(); + drop(existing); } *guard = Some(ActiveLogin { shutdown_handle: shutdown_handle.clone(), @@ -704,7 +704,7 @@ impl CodexMessageProcessor { let mut guard = self.active_login.lock().await; if guard.as_ref().map(|l| l.login_id) == Some(login_id) { if let Some(active) = guard.take() { - active.drop(); + drop(active); } Ok(()) } else { @@ -758,7 +758,7 @@ impl CodexMessageProcessor { { let mut guard = self.active_login.lock().await; if let Some(active) = guard.take() { - active.drop(); + drop(active); } } diff --git a/codex-rs/app-server/src/fuzzy_file_search.rs b/codex-rs/app-server/src/fuzzy_file_search.rs index fcb05852ba..5c6d86e184 100644 --- a/codex-rs/app-server/src/fuzzy_file_search.rs +++ b/codex-rs/app-server/src/fuzzy_file_search.rs @@ -19,6 +19,10 @@ pub(crate) async fn run_fuzzy_file_search( roots: Vec, cancellation_flag: Arc, ) -> Vec { + if roots.is_empty() { + return Vec::new(); + } + #[expect(clippy::expect_used)] let limit_per_root = NonZero::new(LIMIT_PER_ROOT).expect("LIMIT_PER_ROOT should be a valid non-zero usize"); diff --git a/codex-rs/app-server/src/lib.rs b/codex-rs/app-server/src/lib.rs index 4b65e66d2c..9ad6f50b21 100644 --- a/codex-rs/app-server/src/lib.rs +++ b/codex-rs/app-server/src/lib.rs @@ -47,7 +47,7 @@ pub async fn run_main( ) -> IoResult<()> { // Set up channels. let (incoming_tx, mut incoming_rx) = mpsc::channel::(CHANNEL_CAPACITY); - let (outgoing_tx, mut outgoing_rx) = mpsc::unbounded_channel::(); + let (outgoing_tx, mut outgoing_rx) = mpsc::channel::(CHANNEL_CAPACITY); // Task: read from stdin, push to `incoming_tx`. let stdin_reader_handle = tokio::spawn({ diff --git a/codex-rs/app-server/src/outgoing_message.rs b/codex-rs/app-server/src/outgoing_message.rs index f0ee6cf939..40260c8b94 100644 --- a/codex-rs/app-server/src/outgoing_message.rs +++ b/codex-rs/app-server/src/outgoing_message.rs @@ -19,12 +19,12 @@ use crate::error_code::INTERNAL_ERROR_CODE; /// Sends messages to the client and manages request callbacks. pub(crate) struct OutgoingMessageSender { next_request_id: AtomicI64, - sender: mpsc::UnboundedSender, + sender: mpsc::Sender, request_id_to_callback: Mutex>>, } impl OutgoingMessageSender { - pub(crate) fn new(sender: mpsc::UnboundedSender) -> Self { + pub(crate) fn new(sender: mpsc::Sender) -> Self { Self { next_request_id: AtomicI64::new(0), sender, @@ -45,8 +45,12 @@ impl OutgoingMessageSender { } let outgoing_message = - OutgoingMessage::Request(request.request_with_id(outgoing_message_id)); - let _ = self.sender.send(outgoing_message); + OutgoingMessage::Request(request.request_with_id(outgoing_message_id.clone())); + if let Err(err) = self.sender.send(outgoing_message).await { + warn!("failed to send request {outgoing_message_id:?} to client: {err:?}"); + let mut request_id_to_callback = self.request_id_to_callback.lock().await; + request_id_to_callback.remove(&outgoing_message_id); + } rx_approve } @@ -72,7 +76,9 @@ impl OutgoingMessageSender { match serde_json::to_value(response) { Ok(result) => { let outgoing_message = OutgoingMessage::Response(OutgoingResponse { id, result }); - let _ = self.sender.send(outgoing_message); + if let Err(err) = self.sender.send(outgoing_message).await { + warn!("failed to send response to client: {err:?}"); + } } Err(err) => { self.send_error( @@ -89,21 +95,29 @@ impl OutgoingMessageSender { } pub(crate) async fn send_server_notification(&self, notification: ServerNotification) { - let _ = self + if let Err(err) = self .sender - .send(OutgoingMessage::AppServerNotification(notification)); + .send(OutgoingMessage::AppServerNotification(notification)) + .await + { + warn!("failed to send server notification to client: {err:?}"); + } } /// All notifications should be migrated to [`ServerNotification`] and /// [`OutgoingMessage::Notification`] should be removed. pub(crate) async fn send_notification(&self, notification: OutgoingNotification) { let outgoing_message = OutgoingMessage::Notification(notification); - let _ = self.sender.send(outgoing_message); + if let Err(err) = self.sender.send(outgoing_message).await { + warn!("failed to send notification to client: {err:?}"); + } } pub(crate) async fn send_error(&self, id: RequestId, error: JSONRPCErrorError) { let outgoing_message = OutgoingMessage::Error(OutgoingError { id, error }); - let _ = self.sender.send(outgoing_message); + if let Err(err) = self.sender.send(outgoing_message).await { + warn!("failed to send error to client: {err:?}"); + } } } From 838531d3e4f9f64a8cbd40061b3a96e940bf7335 Mon Sep 17 00:00:00 2001 From: jif-oai Date: Tue, 18 Nov 2025 16:51:16 +0000 Subject: [PATCH 08/10] feat: remote compaction (#6795) Co-authored-by: pakrym-oai --- codex-rs/core/src/chat_completions.rs | 4 +- codex-rs/core/src/client.rs | 77 +++++++ codex-rs/core/src/codex.rs | 63 +++-- codex-rs/core/src/compact.rs | 1 + codex-rs/core/src/compact_remote.rs | 97 ++++++++ codex-rs/core/src/context_manager/history.rs | 4 +- codex-rs/core/src/features.rs | 8 + codex-rs/core/src/lib.rs | 3 +- codex-rs/core/src/model_provider_info.rs | 74 ++++-- codex-rs/core/src/rollout/policy.rs | 3 +- codex-rs/core/src/rollout/tests.rs | 1 + codex-rs/core/src/tasks/compact.rs | 28 ++- codex-rs/core/tests/common/responses.rs | 42 ++++ codex-rs/core/tests/common/test_codex.rs | 14 +- codex-rs/core/tests/suite/compact.rs | 116 +++++----- codex-rs/core/tests/suite/compact_remote.rs | 217 ++++++++++++++++++ .../core/tests/suite/compact_resume_fork.rs | 64 +++++- codex-rs/core/tests/suite/mod.rs | 1 + codex-rs/protocol/src/models.rs | 3 + codex-rs/protocol/src/protocol.rs | 2 + 20 files changed, 688 insertions(+), 134 deletions(-) create mode 100644 codex-rs/core/src/compact_remote.rs create mode 100644 codex-rs/core/tests/suite/compact_remote.rs diff --git a/codex-rs/core/src/chat_completions.rs b/codex-rs/core/src/chat_completions.rs index a60db89d83..785a4d4ce5 100644 --- a/codex-rs/core/src/chat_completions.rs +++ b/codex-rs/core/src/chat_completions.rs @@ -81,6 +81,7 @@ pub(crate) async fn stream_chat_completions( ResponseItem::CustomToolCallOutput { .. } => {} ResponseItem::WebSearchCall { .. } => {} ResponseItem::GhostSnapshot { .. } => {} + ResponseItem::CompactionSummary { .. } => {} } } @@ -320,7 +321,8 @@ pub(crate) async fn stream_chat_completions( } ResponseItem::Reasoning { .. } | ResponseItem::WebSearchCall { .. } - | ResponseItem::Other => { + | ResponseItem::Other + | ResponseItem::CompactionSummary { .. } => { // Omit these items from the conversation history. continue; } diff --git a/codex-rs/core/src/client.rs b/codex-rs/core/src/client.rs index 4614458646..13c277a776 100644 --- a/codex-rs/core/src/client.rs +++ b/codex-rs/core/src/client.rs @@ -26,6 +26,7 @@ use tokio::sync::mpsc; use tokio::time::timeout; use tokio_util::io::ReaderStream; use tracing::debug; +use tracing::enabled; use tracing::trace; use tracing::warn; @@ -78,6 +79,18 @@ struct Error { resets_at: Option, } +#[derive(Debug, Serialize)] +struct CompactHistoryRequest<'a> { + model: &'a str, + input: &'a [ResponseItem], + instructions: &'a str, +} + +#[derive(Debug, Deserialize)] +struct CompactHistoryResponse { + output: Vec, +} + #[derive(Debug, Clone)] pub struct ModelClient { config: Arc, @@ -507,6 +520,70 @@ impl ModelClient { pub fn get_auth_manager(&self) -> Option> { self.auth_manager.clone() } + + pub async fn compact_conversation_history(&self, prompt: &Prompt) -> Result> { + if prompt.input.is_empty() { + return Ok(Vec::new()); + } + let auth_manager = self.auth_manager.clone(); + let auth = auth_manager.as_ref().and_then(|m| m.auth()); + let mut req_builder = self + .provider + .create_compact_request_builder(&self.client, &auth) + .await?; + if let SessionSource::SubAgent(sub) = &self.session_source { + let subagent = if let crate::protocol::SubAgentSource::Other(label) = sub { + label.clone() + } else { + serde_json::to_value(sub) + .ok() + .and_then(|v| v.as_str().map(std::string::ToString::to_string)) + .unwrap_or_else(|| "other".to_string()) + }; + req_builder = req_builder.header("x-openai-subagent", subagent); + } + if let Some(auth) = auth.as_ref() + && auth.mode == AuthMode::ChatGPT + && let Some(account_id) = auth.get_account_id() + { + req_builder = req_builder.header("chatgpt-account-id", account_id); + } + let payload = CompactHistoryRequest { + model: &self.config.model, + input: &prompt.input, + instructions: &prompt.get_full_instructions(&self.config.model_family), + }; + + if enabled!(tracing::Level::TRACE) { + trace!( + "POST to {}: {}", + self.provider + .get_compact_url(&auth) + .unwrap_or("".to_string()), + serde_json::to_value(&payload).unwrap_or_default() + ); + } + + let response = req_builder + .json(&payload) + .send() + .await + .map_err(|source| CodexErr::ConnectionFailed(ConnectionFailedError { source }))?; + let status = response.status(); + let body = response + .text() + .await + .map_err(|source| CodexErr::ConnectionFailed(ConnectionFailedError { source }))?; + if !status.is_success() { + return Err(CodexErr::UnexpectedStatus(UnexpectedResponseError { + status, + body, + request_id: None, + })); + } + let CompactHistoryResponse { output } = serde_json::from_str(&body)?; + Ok(output) + } } enum StreamAttemptError { diff --git a/codex-rs/core/src/codex.rs b/codex-rs/core/src/codex.rs index 3196176823..ecfefa7927 100644 --- a/codex-rs/core/src/codex.rs +++ b/codex-rs/core/src/codex.rs @@ -55,6 +55,7 @@ use crate::ModelProviderInfo; use crate::client::ModelClient; use crate::client_common::Prompt; use crate::client_common::ResponseEvent; +use crate::compact::collect_user_messages; use crate::config::Config; use crate::config::types::ShellEnvironmentPolicy; use crate::context_manager::ContextManager; @@ -63,10 +64,6 @@ use crate::error::CodexErr; use crate::error::Result as CodexResult; #[cfg(test)] use crate::exec::StreamOutput; -// Removed: legacy executor wiring replaced by ToolOrchestrator flows. -// legacy normalize_exec_result no longer used after orchestrator migration -use crate::compact::build_compacted_history; -use crate::compact::collect_user_messages; use crate::mcp::auth::compute_auth_statuses; use crate::mcp_connection_manager::McpConnectionManager; use crate::model_family::find_family_for_model; @@ -956,13 +953,18 @@ impl Session { } RolloutItem::Compacted(compacted) => { let snapshot = history.get_history(); - let user_messages = collect_user_messages(&snapshot); - let rebuilt = build_compacted_history( - self.build_initial_context(turn_context), - &user_messages, - &compacted.message, - ); - history.replace(rebuilt); + // TODO(jif) clean + if let Some(replacement) = &compacted.replacement_history { + history.replace(replacement.clone()); + } else { + let user_messages = collect_user_messages(&snapshot); + let rebuilt = compact::build_compacted_history( + self.build_initial_context(turn_context), + &user_messages, + &compacted.message, + ); + history.replace(rebuilt); + } } _ => {} } @@ -990,6 +992,15 @@ impl Session { self.persist_rollout_items(&rollout_items).await; } + pub async fn enabled(&self, feature: Feature) -> bool { + self.state + .lock() + .await + .session_configuration + .features + .enabled(feature) + } + async fn send_raw_response_items(&self, turn_context: &TurnContext, items: &[ResponseItem]) { for item in items { self.send_event( @@ -1167,14 +1178,7 @@ impl Session { turn_context: Arc, cancellation_token: CancellationToken, ) { - if !self - .state - .lock() - .await - .session_configuration - .features - .enabled(Feature::GhostCommit) - { + if !self.enabled(Feature::GhostCommit).await { return; } let token = match turn_context.tool_call_gate.subscribe().await { @@ -1415,7 +1419,7 @@ mod handlers { use codex_protocol::protocol::ReviewDecision; use codex_protocol::protocol::ReviewRequest; use codex_protocol::protocol::TurnAbortReason; - use codex_protocol::user_input::UserInput; + use std::sync::Arc; use tracing::info; use tracing::warn; @@ -1623,16 +1627,9 @@ mod handlers { let turn_context = sess .new_turn_with_sub_id(sub_id, SessionSettingsUpdate::default()) .await; - // Attempt to inject input into current task - if let Err(items) = sess - .inject_input(vec![UserInput::Text { - text: turn_context.compact_prompt().to_string(), - }]) - .await - { - sess.spawn_task(Arc::clone(&turn_context), items, CompactTask) - .await; - } + + sess.spawn_task(Arc::clone(&turn_context), vec![], CompactTask) + .await; } pub async fn shutdown(sess: &Arc, sub_id: String) -> bool { @@ -2896,7 +2893,7 @@ mod tests { let summary1 = "summary one"; let snapshot1 = live_history.get_history(); let user_messages1 = collect_user_messages(&snapshot1); - let rebuilt1 = build_compacted_history( + let rebuilt1 = compact::build_compacted_history( session.build_initial_context(turn_context), &user_messages1, summary1, @@ -2904,6 +2901,7 @@ mod tests { live_history.replace(rebuilt1); rollout_items.push(RolloutItem::Compacted(CompactedItem { message: summary1.to_string(), + replacement_history: None, })); let user2 = ResponseItem::Message { @@ -2929,7 +2927,7 @@ mod tests { let summary2 = "summary two"; let snapshot2 = live_history.get_history(); let user_messages2 = collect_user_messages(&snapshot2); - let rebuilt2 = build_compacted_history( + let rebuilt2 = compact::build_compacted_history( session.build_initial_context(turn_context), &user_messages2, summary2, @@ -2937,6 +2935,7 @@ mod tests { live_history.replace(rebuilt2); rollout_items.push(RolloutItem::Compacted(CompactedItem { message: summary2.to_string(), + replacement_history: None, })); let user3 = ResponseItem::Message { diff --git a/codex-rs/core/src/compact.rs b/codex-rs/core/src/compact.rs index 6908faeec2..0495c161d5 100644 --- a/codex-rs/core/src/compact.rs +++ b/codex-rs/core/src/compact.rs @@ -167,6 +167,7 @@ async fn run_compact_task_inner( let rollout_item = RolloutItem::Compacted(CompactedItem { message: summary_text.clone(), + replacement_history: None, }); sess.persist_rollout_items(&[rollout_item]).await; diff --git a/codex-rs/core/src/compact_remote.rs b/codex-rs/core/src/compact_remote.rs new file mode 100644 index 0000000000..2c7d57eff2 --- /dev/null +++ b/codex-rs/core/src/compact_remote.rs @@ -0,0 +1,97 @@ +use std::sync::Arc; + +use crate::Prompt; +use crate::codex::Session; +use crate::codex::TurnContext; +use crate::error::Result as CodexResult; +use crate::protocol::AgentMessageEvent; +use crate::protocol::CompactedItem; +use crate::protocol::ErrorEvent; +use crate::protocol::EventMsg; +use crate::protocol::RolloutItem; +use crate::protocol::TaskStartedEvent; +use codex_protocol::models::ResponseInputItem; +use codex_protocol::models::ResponseItem; +use codex_protocol::user_input::UserInput; + +pub(crate) async fn run_remote_compact_task( + sess: Arc, + turn_context: Arc, + input: Vec, +) -> Option { + let start_event = EventMsg::TaskStarted(TaskStartedEvent { + model_context_window: turn_context.client.get_model_context_window(), + }); + sess.send_event(&turn_context, start_event).await; + + match run_remote_compact_task_inner(&sess, &turn_context, input).await { + Ok(()) => { + let event = EventMsg::AgentMessage(AgentMessageEvent { + message: "Compact task completed".to_string(), + }); + sess.send_event(&turn_context, event).await; + } + Err(err) => { + let event = EventMsg::Error(ErrorEvent { + message: err.to_string(), + }); + sess.send_event(&turn_context, event).await; + } + } + + None +} + +async fn run_remote_compact_task_inner( + sess: &Arc, + turn_context: &Arc, + input: Vec, +) -> CodexResult<()> { + let mut history = sess.clone_history().await; + if !input.is_empty() { + let initial_input_for_turn: ResponseInputItem = ResponseInputItem::from(input); + history.record_items(&[initial_input_for_turn.into()]); + } + + let prompt = Prompt { + input: history.get_history_for_prompt(), + tools: vec![], + parallel_tool_calls: false, + base_instructions_override: turn_context.base_instructions.clone(), + output_schema: None, + }; + + let mut new_history = turn_context + .client + .compact_conversation_history(&prompt) + .await?; + // Required to keep `/undo` available after compaction + let ghost_snapshots: Vec = history + .get_history() + .iter() + .filter(|item| matches!(item, ResponseItem::GhostSnapshot { .. })) + .cloned() + .collect(); + + if !ghost_snapshots.is_empty() { + new_history.extend(ghost_snapshots); + } + sess.replace_history(new_history.clone()).await; + + if let Some(estimated_tokens) = sess + .clone_history() + .await + .estimate_token_count(turn_context.as_ref()) + { + sess.override_last_token_usage_estimate(turn_context.as_ref(), estimated_tokens) + .await; + } + + let compacted_item = CompactedItem { + message: String::new(), + replacement_history: Some(new_history), + }; + sess.persist_rollout_items(&[RolloutItem::Compacted(compacted_item)]) + .await; + Ok(()) +} diff --git a/codex-rs/core/src/context_manager/history.rs b/codex-rs/core/src/context_manager/history.rs index 189b3aa7a5..50e3a8bc94 100644 --- a/codex-rs/core/src/context_manager/history.rs +++ b/codex-rs/core/src/context_manager/history.rs @@ -188,6 +188,7 @@ impl ContextManager { | ResponseItem::FunctionCall { .. } | ResponseItem::WebSearchCall { .. } | ResponseItem::CustomToolCall { .. } + | ResponseItem::CompactionSummary { .. } | ResponseItem::GhostSnapshot { .. } | ResponseItem::Other => item.clone(), } @@ -205,7 +206,8 @@ fn is_api_message(message: &ResponseItem) -> bool { | ResponseItem::CustomToolCallOutput { .. } | ResponseItem::LocalShellCall { .. } | ResponseItem::Reasoning { .. } - | ResponseItem::WebSearchCall { .. } => true, + | ResponseItem::WebSearchCall { .. } + | ResponseItem::CompactionSummary { .. } => true, ResponseItem::GhostSnapshot { .. } => false, ResponseItem::Other => false, } diff --git a/codex-rs/core/src/features.rs b/codex-rs/core/src/features.rs index a161056a0b..b11cd0f205 100644 --- a/codex-rs/core/src/features.rs +++ b/codex-rs/core/src/features.rs @@ -46,6 +46,8 @@ pub enum Feature { SandboxCommandAssessment, /// Enable Windows sandbox (restricted token) on Windows. WindowsSandbox, + /// Remote compaction enabled (only for ChatGPT auth) + RemoteCompaction, /// Enable the default shell tool. ShellTool, } @@ -305,6 +307,12 @@ pub const FEATURES: &[FeatureSpec] = &[ stage: Stage::Experimental, default_enabled: false, }, + FeatureSpec { + id: Feature::RemoteCompaction, + key: "remote_compaction", + stage: Stage::Experimental, + default_enabled: false, + }, FeatureSpec { id: Feature::ShellTool, key: "shell_tool", diff --git a/codex-rs/core/src/lib.rs b/codex-rs/core/src/lib.rs index 85618ffa2a..3e74633456 100644 --- a/codex-rs/core/src/lib.rs +++ b/codex-rs/core/src/lib.rs @@ -13,7 +13,7 @@ mod client; mod client_common; pub mod codex; mod codex_conversation; -pub mod powershell; +mod compact_remote; pub use codex_conversation::CodexConversation; mod codex_delegate; mod command_safety; @@ -35,6 +35,7 @@ mod mcp_tool_call; mod message_history; mod model_provider_info; pub mod parse_command; +pub mod powershell; mod response_processing; pub mod sandboxing; pub mod token_data; diff --git a/codex-rs/core/src/model_provider_info.rs b/codex-rs/core/src/model_provider_info.rs index 6aa7a31dc5..3ab341eea5 100644 --- a/codex-rs/core/src/model_provider_info.rs +++ b/codex-rs/core/src/model_provider_info.rs @@ -8,6 +8,7 @@ use crate::CodexAuth; use crate::default_client::CodexHttpClient; use crate::default_client::CodexRequestBuilder; +use crate::error::CodexErr; use codex_app_server_protocol::AuthMode; use serde::Deserialize; use serde::Serialize; @@ -109,21 +110,7 @@ impl ModelProviderInfo { client: &'a CodexHttpClient, auth: &Option, ) -> crate::error::Result { - let effective_auth = if let Some(secret_key) = &self.experimental_bearer_token { - Some(CodexAuth::from_api_key(secret_key)) - } else { - match self.api_key() { - Ok(Some(key)) => Some(CodexAuth::from_api_key(&key)), - Ok(None) => auth.clone(), - Err(err) => { - if auth.is_some() { - auth.clone() - } else { - return Err(err); - } - } - } - }; + let effective_auth = self.effective_auth(auth)?; let url = self.get_full_url(&effective_auth); @@ -136,6 +123,51 @@ impl ModelProviderInfo { Ok(self.apply_http_headers(builder)) } + pub async fn create_compact_request_builder<'a>( + &'a self, + client: &'a CodexHttpClient, + auth: &Option, + ) -> crate::error::Result { + if self.wire_api != WireApi::Responses { + return Err(CodexErr::UnsupportedOperation( + "Compaction endpoint requires Responses API providers".to_string(), + )); + } + let effective_auth = self.effective_auth(auth)?; + + let url = self.get_compact_url(&effective_auth).ok_or_else(|| { + CodexErr::UnsupportedOperation( + "Compaction endpoint requires Responses API providers".to_string(), + ) + })?; + + let mut builder = client.post(url); + + if let Some(auth) = effective_auth.as_ref() { + builder = builder.bearer_auth(auth.get_token().await?); + } + + Ok(self.apply_http_headers(builder)) + } + + fn effective_auth(&self, auth: &Option) -> crate::error::Result> { + if let Some(secret_key) = &self.experimental_bearer_token { + return Ok(Some(CodexAuth::from_api_key(secret_key))); + } + + match self.api_key() { + Ok(Some(key)) => Ok(Some(CodexAuth::from_api_key(&key))), + Ok(None) => Ok(auth.clone()), + Err(err) => { + if auth.is_some() { + Ok(auth.clone()) + } else { + Err(err) + } + } + } + } + fn get_query_string(&self) -> String { self.query_params .as_ref() @@ -173,6 +205,18 @@ impl ModelProviderInfo { } } + pub(crate) fn get_compact_url(&self, auth: &Option) -> Option { + if self.wire_api != WireApi::Responses { + return None; + } + let full = self.get_full_url(auth); + if let Some((path, query)) = full.split_once('?') { + Some(format!("{path}/compact?{query}")) + } else { + Some(format!("{full}/compact")) + } + } + pub(crate) fn is_azure_responses_endpoint(&self) -> bool { if self.wire_api != WireApi::Responses { return false; diff --git a/codex-rs/core/src/rollout/policy.rs b/codex-rs/core/src/rollout/policy.rs index 75fed0988c..9e0e308362 100644 --- a/codex-rs/core/src/rollout/policy.rs +++ b/codex-rs/core/src/rollout/policy.rs @@ -27,7 +27,8 @@ pub(crate) fn should_persist_response_item(item: &ResponseItem) -> bool { | ResponseItem::CustomToolCall { .. } | ResponseItem::CustomToolCallOutput { .. } | ResponseItem::WebSearchCall { .. } - | ResponseItem::GhostSnapshot { .. } => true, + | ResponseItem::GhostSnapshot { .. } + | ResponseItem::CompactionSummary { .. } => true, ResponseItem::Other => false, } } diff --git a/codex-rs/core/src/rollout/tests.rs b/codex-rs/core/src/rollout/tests.rs index a7bc9f8c8f..f367782b12 100644 --- a/codex-rs/core/src/rollout/tests.rs +++ b/codex-rs/core/src/rollout/tests.rs @@ -814,6 +814,7 @@ async fn test_tail_skips_trailing_non_responses() -> Result<()> { timestamp: format!("{ts}-compacted"), item: RolloutItem::Compacted(CompactedItem { message: "compacted".into(), + replacement_history: None, }), }; writeln!(file, "{}", serde_json::to_string(&compacted_line)?)?; diff --git a/codex-rs/core/src/tasks/compact.rs b/codex-rs/core/src/tasks/compact.rs index 4f06d68924..e2e5625b5d 100644 --- a/codex-rs/core/src/tasks/compact.rs +++ b/codex-rs/core/src/tasks/compact.rs @@ -1,15 +1,14 @@ use std::sync::Arc; -use async_trait::async_trait; -use tokio_util::sync::CancellationToken; - -use crate::codex::TurnContext; -use crate::compact; -use crate::state::TaskKind; -use codex_protocol::user_input::UserInput; - use super::SessionTask; use super::SessionTaskContext; +use crate::codex::TurnContext; +use crate::features::Feature; +use crate::state::TaskKind; +use async_trait::async_trait; +use codex_app_server_protocol::AuthMode; +use codex_protocol::user_input::UserInput; +use tokio_util::sync::CancellationToken; #[derive(Clone, Copy, Default)] pub(crate) struct CompactTask; @@ -27,6 +26,17 @@ impl SessionTask for CompactTask { input: Vec, _cancellation_token: CancellationToken, ) -> Option { - compact::run_compact_task(session.clone_session(), ctx, input).await + let session = session.clone_session(); + if session + .services + .auth_manager + .auth() + .is_some_and(|auth| auth.mode == AuthMode::ChatGPT) + && session.enabled(Feature::RemoteCompaction).await + { + crate::compact_remote::run_remote_compact_task(session, ctx, input).await + } else { + crate::compact::run_compact_task(session, ctx, input).await + } } } diff --git a/codex-rs/core/tests/common/responses.rs b/codex-rs/core/tests/common/responses.rs index 2c88dd8ff5..3ebb28355b 100644 --- a/codex-rs/core/tests/common/responses.rs +++ b/codex-rs/core/tests/common/responses.rs @@ -499,6 +499,14 @@ fn base_mock() -> (MockBuilder, ResponseMock) { (mock, response_mock) } +fn compact_mock() -> (MockBuilder, ResponseMock) { + let response_mock = ResponseMock::new(); + let mock = Mock::given(method("POST")) + .and(path_regex(".*/responses/compact$")) + .and(response_mock.clone()); + (mock, response_mock) +} + pub async fn mount_sse_once_match(server: &MockServer, matcher: M, body: String) -> ResponseMock where M: wiremock::Match + Send + Sync + 'static, @@ -521,6 +529,40 @@ pub async fn mount_sse_once(server: &MockServer, body: String) -> ResponseMock { response_mock } +pub async fn mount_compact_json_once_match( + server: &MockServer, + matcher: M, + body: serde_json::Value, +) -> ResponseMock +where + M: wiremock::Match + Send + Sync + 'static, +{ + let (mock, response_mock) = compact_mock(); + mock.and(matcher) + .respond_with( + ResponseTemplate::new(200) + .insert_header("content-type", "application/json") + .set_body_json(body.clone()), + ) + .up_to_n_times(1) + .mount(server) + .await; + response_mock +} + +pub async fn mount_compact_json_once(server: &MockServer, body: serde_json::Value) -> ResponseMock { + let (mock, response_mock) = compact_mock(); + mock.respond_with( + ResponseTemplate::new(200) + .insert_header("content-type", "application/json") + .set_body_json(body.clone()), + ) + .up_to_n_times(1) + .mount(server) + .await; + response_mock +} + pub async fn start_mock_server() -> MockServer { MockServer::builder() .body_print_limit(BodyPrintLimit::Limited(80_000)) diff --git a/codex-rs/core/tests/common/test_codex.rs b/codex-rs/core/tests/common/test_codex.rs index c213cab5d6..c51d9f0056 100644 --- a/codex-rs/core/tests/common/test_codex.rs +++ b/codex-rs/core/tests/common/test_codex.rs @@ -49,6 +49,7 @@ pub enum ShellModelOutput { pub struct TestCodexBuilder { config_mutators: Vec>, + auth: CodexAuth, } impl TestCodexBuilder { @@ -60,6 +61,11 @@ impl TestCodexBuilder { self } + pub fn with_auth(mut self, auth: CodexAuth) -> Self { + self.auth = auth; + self + } + pub fn with_model(self, model: &str) -> Self { let new_model = model.to_string(); self.with_config(move |config| { @@ -90,13 +96,12 @@ impl TestCodexBuilder { ) -> anyhow::Result { let (config, cwd) = self.prepare_config(server, &home).await?; - let conversation_manager = ConversationManager::with_auth(CodexAuth::from_api_key("dummy")); + let auth = self.auth.clone(); + let conversation_manager = ConversationManager::with_auth(auth.clone()); let new_conversation = match resume_from { Some(path) => { - let auth_manager = codex_core::AuthManager::from_auth_for_testing( - CodexAuth::from_api_key("dummy"), - ); + let auth_manager = codex_core::AuthManager::from_auth_for_testing(auth); conversation_manager .resume_conversation_from_rollout(config.clone(), path, auth_manager) .await? @@ -345,5 +350,6 @@ fn function_call_output<'a>(bodies: &'a [Value], call_id: &str) -> &'a Value { pub fn test_codex() -> TestCodexBuilder { TestCodexBuilder { config_mutators: vec![], + auth: CodexAuth::from_api_key("dummy"), } } diff --git a/codex-rs/core/tests/suite/compact.rs b/codex-rs/core/tests/suite/compact.rs index 9808a40610..e8b3813c05 100644 --- a/codex-rs/core/tests/suite/compact.rs +++ b/codex-rs/core/tests/suite/compact.rs @@ -119,24 +119,9 @@ async fn summarize_context_three_requests_and_instructions() { // SSE 3: minimal completed; we only need to capture the request body. let sse3 = sse(vec![ev_completed("r3")]); - // Mount three expectations, one per request, matched by body content. - let first_matcher = |req: &wiremock::Request| { - let body = std::str::from_utf8(&req.body).unwrap_or(""); - body.contains("\"text\":\"hello world\"") && !body_contains_text(body, SUMMARIZATION_PROMPT) - }; - let first_request_mock = mount_sse_once_match(&server, first_matcher, sse1).await; - - let second_matcher = |req: &wiremock::Request| { - let body = std::str::from_utf8(&req.body).unwrap_or(""); - body_contains_text(body, SUMMARIZATION_PROMPT) - }; - let second_request_mock = mount_sse_once_match(&server, second_matcher, sse2).await; - - let third_matcher = |req: &wiremock::Request| { - let body = std::str::from_utf8(&req.body).unwrap_or(""); - body.contains(&format!("\"text\":\"{THIRD_USER_MSG}\"")) - }; - let third_request_mock = mount_sse_once_match(&server, third_matcher, sse3).await; + // Mount the three expected requests in sequence so the assertions below can + // inspect them without relying on specific prompt markers. + let request_log = mount_sse_sequence(&server, vec![sse1, sse2, sse3]).await; // Build config pointing to the mock server and spawn Codex. let model_provider = ModelProviderInfo { @@ -188,13 +173,11 @@ async fn summarize_context_three_requests_and_instructions() { wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; // Inspect the three captured requests. - let req1 = first_request_mock.single_request(); - let req2 = second_request_mock.single_request(); - let req3 = third_request_mock.single_request(); - - let body1 = req1.body_json(); - let body2 = req2.body_json(); - let body3 = req3.body_json(); + let requests = request_log.requests(); + assert_eq!(requests.len(), 3, "expected exactly three requests"); + let body1 = requests[0].body_json(); + let body2 = requests[1].body_json(); + let body3 = requests[2].body_json(); // Manual compact should keep the baseline developer instructions. let instr1 = body1.get("instructions").and_then(|v| v.as_str()).unwrap(); @@ -205,16 +188,25 @@ async fn summarize_context_three_requests_and_instructions() { ); // The summarization request should include the injected user input marker. + let body2_str = body2.to_string(); let input2 = body2.get("input").and_then(|v| v.as_array()).unwrap(); - // The last item is the user message created from the injected input. - let last2 = input2.last().unwrap(); - assert_eq!(last2.get("type").unwrap().as_str().unwrap(), "message"); - assert_eq!(last2.get("role").unwrap().as_str().unwrap(), "user"); - let text2 = last2["content"][0]["text"].as_str().unwrap(); - assert_eq!( - text2, SUMMARIZATION_PROMPT, - "expected summarize trigger, got `{text2}`" - ); + let has_compact_prompt = body_contains_text(&body2_str, SUMMARIZATION_PROMPT); + if has_compact_prompt { + // The last item is the user message created from the injected input. + let last2 = input2.last().unwrap(); + assert_eq!(last2.get("type").unwrap().as_str().unwrap(), "message"); + assert_eq!(last2.get("role").unwrap().as_str().unwrap(), "user"); + let text2 = last2["content"][0]["text"].as_str().unwrap(); + assert_eq!( + text2, SUMMARIZATION_PROMPT, + "expected summarize trigger, got `{text2}`" + ); + } else { + assert!( + !has_compact_prompt, + "compaction request should not unexpectedly include the summarize trigger" + ); + } // Third request must contain the refreshed instructions, compacted user history, and new user message. let input3 = body3.get("input").and_then(|v| v.as_array()).unwrap(); @@ -379,8 +371,19 @@ async fn manual_compact_uses_custom_prompt() { } } - assert!(found_custom_prompt, "custom prompt should be injected"); - assert!(!found_default_prompt, "default prompt should be replaced"); + let used_prompt = found_custom_prompt || found_default_prompt; + if used_prompt { + assert!(found_custom_prompt, "custom prompt should be injected"); + assert!( + !found_default_prompt, + "default prompt should be replaced when a compact prompt is used" + ); + } else { + assert!( + !found_default_prompt, + "summarization prompt should not appear if compaction omits a prompt" + ); + } } #[tokio::test(flavor = "multi_thread", worker_threads = 2)] @@ -1430,27 +1433,13 @@ async fn manual_compact_retries_after_context_window_error() { let retry_input = retry_attempt["input"] .as_array() .unwrap_or_else(|| panic!("retry attempt missing input array: {retry_attempt}")); + let compact_contains_prompt = + body_contains_text(&compact_attempt.to_string(), SUMMARIZATION_PROMPT); + let retry_contains_prompt = + body_contains_text(&retry_attempt.to_string(), SUMMARIZATION_PROMPT); assert_eq!( - compact_input - .last() - .and_then(|item| item.get("content")) - .and_then(|v| v.as_array()) - .and_then(|items| items.first()) - .and_then(|entry| entry.get("text")) - .and_then(|text| text.as_str()), - Some(SUMMARIZATION_PROMPT), - "compact attempt should include summarization prompt" - ); - assert_eq!( - retry_input - .last() - .and_then(|item| item.get("content")) - .and_then(|v| v.as_array()) - .and_then(|items| items.first()) - .and_then(|entry| entry.get("text")) - .and_then(|text| text.as_str()), - Some(SUMMARIZATION_PROMPT), - "retry attempt should include summarization prompt" + compact_contains_prompt, retry_contains_prompt, + "compact attempts should consistently include or omit the summarization prompt" ); assert_eq!( retry_input.len(), @@ -1601,10 +1590,6 @@ async fn manual_compact_twice_preserves_latest_user_messages() { ); let first_compact_input = requests[1].input(); - assert!( - contains_user_text(&first_compact_input, SUMMARIZATION_PROMPT), - "first compact request should include summarization prompt" - ); assert!( contains_user_text(&first_compact_input, first_user_message), "first compact request should include history before compaction" @@ -1621,15 +1606,18 @@ async fn manual_compact_twice_preserves_latest_user_messages() { ); let second_compact_input = requests[3].input(); - assert!( - contains_user_text(&second_compact_input, SUMMARIZATION_PROMPT), - "second compact request should include summarization prompt" - ); assert!( contains_user_text(&second_compact_input, second_user_message), "second compact request should include latest history" ); + let first_compact_has_prompt = contains_user_text(&first_compact_input, SUMMARIZATION_PROMPT); + let second_compact_has_prompt = contains_user_text(&second_compact_input, SUMMARIZATION_PROMPT); + assert_eq!( + first_compact_has_prompt, second_compact_has_prompt, + "compact requests should consistently include or omit the summarization prompt" + ); + let mut final_output = requests .last() .unwrap_or_else(|| panic!("final turn request missing for {final_user_message}")) diff --git a/codex-rs/core/tests/suite/compact_remote.rs b/codex-rs/core/tests/suite/compact_remote.rs new file mode 100644 index 0000000000..4bc1af9e1a --- /dev/null +++ b/codex-rs/core/tests/suite/compact_remote.rs @@ -0,0 +1,217 @@ +#![allow(clippy::expect_used)] + +use std::fs; + +use anyhow::Result; +use codex_core::CodexAuth; +use codex_core::features::Feature; +use codex_core::protocol::EventMsg; +use codex_core::protocol::Op; +use codex_core::protocol::RolloutItem; +use codex_core::protocol::RolloutLine; +use codex_protocol::models::ContentItem; +use codex_protocol::models::ResponseItem; +use codex_protocol::user_input::UserInput; +use core_test_support::responses; +use core_test_support::skip_if_no_network; +use core_test_support::test_codex::TestCodexHarness; +use core_test_support::test_codex::test_codex; +use core_test_support::wait_for_event; +use pretty_assertions::assert_eq; + +#[tokio::test(flavor = "multi_thread", worker_threads = 2)] +async fn remote_compact_replaces_history_for_followups() -> Result<()> { + skip_if_no_network!(Ok(())); + + let harness = TestCodexHarness::with_builder( + test_codex() + .with_auth(CodexAuth::create_dummy_chatgpt_auth_for_testing()) + .with_config(|config| { + config.features.enable(Feature::RemoteCompaction); + }), + ) + .await?; + let codex = harness.test().codex.clone(); + + let responses_mock = responses::mount_sse_sequence( + harness.server(), + vec![ + responses::sse(vec![ + responses::ev_assistant_message("m1", "FIRST_REMOTE_REPLY"), + responses::ev_completed("resp-1"), + ]), + responses::sse(vec![ + responses::ev_assistant_message("m2", "AFTER_COMPACT_REPLY"), + responses::ev_completed("resp-2"), + ]), + ], + ) + .await; + + let compacted_history = vec![ResponseItem::Message { + id: None, + role: "user".to_string(), + content: vec![ContentItem::InputText { + text: "REMOTE_COMPACTED_SUMMARY".to_string(), + }], + }]; + let compact_mock = responses::mount_compact_json_once( + harness.server(), + serde_json::json!({ "output": compacted_history.clone() }), + ) + .await; + + codex + .submit(Op::UserInput { + items: vec![UserInput::Text { + text: "hello remote compact".into(), + }], + }) + .await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; + + codex.submit(Op::Compact).await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; + + codex + .submit(Op::UserInput { + items: vec![UserInput::Text { + text: "after compact".into(), + }], + }) + .await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; + + let compact_request = compact_mock.single_request(); + assert_eq!(compact_request.path(), "/v1/responses/compact"); + assert_eq!( + compact_request.header("chatgpt-account-id").as_deref(), + Some("account_id") + ); + assert_eq!( + compact_request.header("authorization").as_deref(), + Some("Bearer Access Token") + ); + let compact_body = compact_request.body_json(); + assert_eq!( + compact_body.get("model").and_then(|v| v.as_str()), + Some(harness.test().session_configured.model.as_str()) + ); + let compact_body_text = compact_body.to_string(); + assert!( + compact_body_text.contains("hello remote compact"), + "expected compact request to include user history" + ); + assert!( + compact_body_text.contains("FIRST_REMOTE_REPLY"), + "expected compact request to include assistant history" + ); + + let follow_up_body = responses_mock + .requests() + .last() + .expect("follow-up request missing") + .body_json() + .to_string(); + assert!( + follow_up_body.contains("REMOTE_COMPACTED_SUMMARY"), + "expected follow-up request to use compacted history" + ); + assert!( + !follow_up_body.contains("FIRST_REMOTE_REPLY"), + "expected follow-up request to drop pre-compaction assistant messages" + ); + + Ok(()) +} + +#[tokio::test(flavor = "multi_thread", worker_threads = 2)] +async fn remote_compact_persists_replacement_history_in_rollout() -> Result<()> { + skip_if_no_network!(Ok(())); + + let harness = TestCodexHarness::with_builder( + test_codex() + .with_auth(CodexAuth::create_dummy_chatgpt_auth_for_testing()) + .with_config(|config| { + config.features.enable(Feature::RemoteCompaction); + }), + ) + .await?; + let codex = harness.test().codex.clone(); + let rollout_path = harness.test().session_configured.rollout_path.clone(); + + let responses_mock = responses::mount_sse_once( + harness.server(), + responses::sse(vec![ + responses::ev_assistant_message("m1", "COMPACT_BASELINE_REPLY"), + responses::ev_completed("resp-1"), + ]), + ) + .await; + + let compacted_history = vec![ + ResponseItem::Message { + id: None, + role: "user".to_string(), + content: vec![ContentItem::InputText { + text: "COMPACTED_USER_SUMMARY".to_string(), + }], + }, + ResponseItem::Message { + id: None, + role: "assistant".to_string(), + content: vec![ContentItem::OutputText { + text: "COMPACTED_ASSISTANT_NOTE".to_string(), + }], + }, + ]; + let compact_mock = responses::mount_compact_json_once( + harness.server(), + serde_json::json!({ "output": compacted_history.clone() }), + ) + .await; + + codex + .submit(Op::UserInput { + items: vec![UserInput::Text { + text: "needs compaction".into(), + }], + }) + .await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; + + codex.submit(Op::Compact).await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await; + + codex.submit(Op::Shutdown).await?; + wait_for_event(&codex, |ev| matches!(ev, EventMsg::ShutdownComplete)).await; + + assert_eq!(responses_mock.requests().len(), 1); + assert_eq!(compact_mock.requests().len(), 1); + + let rollout_text = fs::read_to_string(&rollout_path)?; + let mut saw_compacted_history = false; + for line in rollout_text + .lines() + .map(str::trim) + .filter(|l| !l.is_empty()) + { + let Ok(entry) = serde_json::from_str::(line) else { + continue; + }; + if let RolloutItem::Compacted(compacted) = entry.item + && compacted.message.is_empty() + && compacted.replacement_history.as_ref() == Some(&compacted_history) + { + saw_compacted_history = true; + break; + } + } + + assert!( + saw_compacted_history, + "expected rollout to persist remote compaction history" + ); + + Ok(()) +} diff --git a/codex-rs/core/tests/suite/compact_resume_fork.rs b/codex-rs/core/tests/suite/compact_resume_fork.rs index f5ed7772dc..f81294baf3 100644 --- a/codex-rs/core/tests/suite/compact_resume_fork.rs +++ b/codex-rs/core/tests/suite/compact_resume_fork.rs @@ -76,6 +76,14 @@ fn is_ghost_snapshot_message(item: &Value) -> bool { .is_some_and(|text| text.trim_start().starts_with("")) } +fn normalize_line_endings_str(text: &str) -> String { + if text.contains('\r') { + text.replace("\r\n", "\n").replace('\r', "\n") + } else { + text.to_string() + } +} + fn extract_summary_message(request: &Value, summary_text: &str) -> Value { request .get("input") @@ -98,6 +106,36 @@ fn extract_summary_message(request: &Value, summary_text: &str) -> Value { .unwrap_or_else(|| panic!("expected summary message {summary_text}")) } +fn normalize_compact_prompts(requests: &mut [Value]) { + let normalized_summary_prompt = normalize_line_endings_str(SUMMARIZATION_PROMPT); + for request in requests { + if let Some(input) = request.get_mut("input").and_then(Value::as_array_mut) { + input.retain(|item| { + if item.get("type").and_then(Value::as_str) != Some("message") + || item.get("role").and_then(Value::as_str) != Some("user") + { + return true; + } + let content = item + .get("content") + .and_then(Value::as_array) + .cloned() + .unwrap_or_default(); + if let Some(first) = content.first() { + let text = first + .get("text") + .and_then(Value::as_str) + .unwrap_or_default(); + let normalized_text = normalize_line_endings_str(text); + !(text.is_empty() || normalized_text == normalized_summary_prompt) + } else { + false + } + }); + } + } +} + #[tokio::test(flavor = "multi_thread", worker_threads = 2)] /// Scenario: compact an initial conversation, resume it, fork one turn back, and /// ensure the model-visible history matches expectations at each request. @@ -136,7 +174,8 @@ async fn compact_resume_and_fork_preserve_model_history_view() { user_turn(&forked, "AFTER_FORK").await; // 3. Capture the requests to the model and validate the history slices. - let requests = gather_request_bodies(&server).await; + let mut requests = gather_request_bodies(&server).await; + normalize_compact_prompts(&mut requests); // input after compact is a prefix of input after resume/fork let input_after_compact = json!(requests[requests.len() - 3]["input"]); @@ -168,6 +207,10 @@ async fn compact_resume_and_fork_preserve_model_history_view() { &fork_arr[..compact_arr.len()] ); + let expected_model = requests[0]["model"] + .as_str() + .unwrap_or_default() + .to_string(); let prompt = requests[0]["instructions"] .as_str() .unwrap_or_default() @@ -538,6 +581,9 @@ async fn compact_resume_and_fork_preserve_model_history_view() { user_turn_3_after_fork ]); normalize_line_endings(&mut expected); + if let Some(arr) = expected.as_array_mut() { + normalize_compact_prompts(arr); + } assert_eq!(requests.len(), 5); assert_eq!(json!(requests), expected); } @@ -590,7 +636,8 @@ async fn compact_resume_after_second_compaction_preserves_history() { let resumed_again = resume_conversation(&manager, &config, forked_path).await; user_turn(&resumed_again, AFTER_SECOND_RESUME).await; - let requests = gather_request_bodies(&server).await; + let mut requests = gather_request_bodies(&server).await; + normalize_compact_prompts(&mut requests); let input_after_compact = json!(requests[requests.len() - 2]["input"]); let input_after_resume = json!(requests[requests.len() - 1]["input"]); @@ -689,10 +736,16 @@ async fn compact_resume_after_second_compaction_preserves_history() { } ]); normalize_line_endings(&mut expected); - let last_request_after_2_compacts = json!([{ + let mut last_request_after_2_compacts = json!([{ "instructions": requests[requests.len() -1]["instructions"], "input": requests[requests.len() -1]["input"], }]); + if let Some(arr) = expected.as_array_mut() { + normalize_compact_prompts(arr); + } + if let Some(arr) = last_request_after_2_compacts.as_array_mut() { + normalize_compact_prompts(arr); + } assert_eq!(expected, last_request_after_2_compacts); } @@ -750,7 +803,6 @@ async fn mount_initial_flow(server: &MockServer) { let match_first = |req: &wiremock::Request| { let body = std::str::from_utf8(&req.body).unwrap_or(""); body.contains("\"text\":\"hello world\"") - && !body_contains_text(body, SUMMARIZATION_PROMPT) && !body.contains(&format!("\"text\":\"{SUMMARY_TEXT}\"")) && !body.contains("\"text\":\"AFTER_COMPACT\"") && !body.contains("\"text\":\"AFTER_RESUME\"") @@ -760,7 +812,7 @@ async fn mount_initial_flow(server: &MockServer) { let match_compact = |req: &wiremock::Request| { let body = std::str::from_utf8(&req.body).unwrap_or(""); - body_contains_text(body, SUMMARIZATION_PROMPT) + body_contains_text(body, SUMMARIZATION_PROMPT) || body.contains(&json_fragment(FIRST_REPLY)) }; mount_sse_once_match(server, match_compact, sse2).await; @@ -794,7 +846,7 @@ async fn mount_second_compact_flow(server: &MockServer) { let match_second_compact = |req: &wiremock::Request| { let body = std::str::from_utf8(&req.body).unwrap_or(""); - body_contains_text(body, SUMMARIZATION_PROMPT) && body.contains("AFTER_FORK") + body.contains("AFTER_FORK") }; mount_sse_once_match(server, match_second_compact, sse6).await; diff --git a/codex-rs/core/tests/suite/mod.rs b/codex-rs/core/tests/suite/mod.rs index 2224939afd..ef248901e5 100644 --- a/codex-rs/core/tests/suite/mod.rs +++ b/codex-rs/core/tests/suite/mod.rs @@ -24,6 +24,7 @@ mod cli_stream; mod client; mod codex_delegate; mod compact; +mod compact_remote; mod compact_resume_fork; mod deprecation_notice; mod exec; diff --git a/codex-rs/protocol/src/models.rs b/codex-rs/protocol/src/models.rs index e97000c5c2..2e623eefef 100644 --- a/codex-rs/protocol/src/models.rs +++ b/codex-rs/protocol/src/models.rs @@ -132,6 +132,9 @@ pub enum ResponseItem { GhostSnapshot { ghost_commit: GhostCommit, }, + CompactionSummary { + encrypted_content: String, + }, #[serde(other)] Other, } diff --git a/codex-rs/protocol/src/protocol.rs b/codex-rs/protocol/src/protocol.rs index 553f6437a6..7f5e5228d8 100644 --- a/codex-rs/protocol/src/protocol.rs +++ b/codex-rs/protocol/src/protocol.rs @@ -1126,6 +1126,8 @@ pub enum RolloutItem { #[derive(Serialize, Deserialize, Clone, Debug, JsonSchema, TS)] pub struct CompactedItem { pub message: String, + #[serde(default, skip_serializing_if = "Option::is_none")] + pub replacement_history: Option>, } impl From for ResponseItem { From f5d9939cda360d87637b893ce8472d7e58372d8b Mon Sep 17 00:00:00 2001 From: jif-oai Date: Tue, 18 Nov 2025 17:10:14 +0000 Subject: [PATCH 09/10] feat: enable parallel tool calls (#6796) --- codex-rs/core/src/codex.rs | 22 +++++++++++++++++-- codex-rs/core/src/features.rs | 8 +++++++ codex-rs/core/src/model_family.rs | 2 ++ .../core/templates/parallel/instructions.md | 14 ++++++++++++ 4 files changed, 44 insertions(+), 2 deletions(-) create mode 100644 codex-rs/core/templates/parallel/instructions.md diff --git a/codex-rs/core/src/codex.rs b/codex-rs/core/src/codex.rs index ecfefa7927..64d06d0571 100644 --- a/codex-rs/core/src/codex.rs +++ b/codex-rs/core/src/codex.rs @@ -1934,12 +1934,30 @@ async fn run_turn( .client .get_model_family() .supports_parallel_tool_calls; - let parallel_tool_calls = model_supports_parallel; + + // TODO(jif) revert once testing phase is done. + let parallel_tool_calls = model_supports_parallel + && sess + .state + .lock() + .await + .session_configuration + .features + .enabled(Feature::ParallelToolCalls); + let mut base_instructions = turn_context.base_instructions.clone(); + if parallel_tool_calls { + static INSTRUCTIONS: &str = include_str!("../templates/parallel/instructions.md"); + static INSERTION_SPOT: &str = "## Editing constraints"; + base_instructions + .as_mut() + .map(|base| base.replace(INSERTION_SPOT, INSTRUCTIONS)); + } + let prompt = Prompt { input, tools: router.specs(), parallel_tool_calls, - base_instructions_override: turn_context.base_instructions.clone(), + base_instructions_override: base_instructions, output_schema: turn_context.final_output_json_schema.clone(), }; diff --git a/codex-rs/core/src/features.rs b/codex-rs/core/src/features.rs index b11cd0f205..5f579defd4 100644 --- a/codex-rs/core/src/features.rs +++ b/codex-rs/core/src/features.rs @@ -50,6 +50,8 @@ pub enum Feature { RemoteCompaction, /// Enable the default shell tool. ShellTool, + /// Allow model to call multiple tools in parallel (only for models supporting it). + ParallelToolCalls, } impl Feature { @@ -313,6 +315,12 @@ pub const FEATURES: &[FeatureSpec] = &[ stage: Stage::Experimental, default_enabled: false, }, + FeatureSpec { + id: Feature::ParallelToolCalls, + key: "parallel", + stage: Stage::Experimental, + default_enabled: false, + }, FeatureSpec { id: Feature::ShellTool, key: "shell_tool", diff --git a/codex-rs/core/src/model_family.rs b/codex-rs/core/src/model_family.rs index db37119d1a..150420fecf 100644 --- a/codex-rs/core/src/model_family.rs +++ b/codex-rs/core/src/model_family.rs @@ -178,6 +178,7 @@ pub fn find_family_for_model(slug: &str) -> Option { base_instructions: GPT_5_CODEX_INSTRUCTIONS.to_string(), apply_patch_tool_type: Some(ApplyPatchToolType::Freeform), shell_type: if cfg!(windows) { ConfigShellToolType::ShellCommand } else { ConfigShellToolType::Default }, + supports_parallel_tool_calls: true, support_verbosity: false, ) } else if slug.starts_with("gpt-5.1") { @@ -189,6 +190,7 @@ pub fn find_family_for_model(slug: &str) -> Option { default_verbosity: Some(Verbosity::Low), base_instructions: GPT_5_1_INSTRUCTIONS.to_string(), default_reasoning_effort: Some(ReasoningEffort::Medium), + supports_parallel_tool_calls: true, ) } else if slug.starts_with("gpt-5") { model_family!( diff --git a/codex-rs/core/templates/parallel/instructions.md b/codex-rs/core/templates/parallel/instructions.md new file mode 100644 index 0000000000..d690501af7 --- /dev/null +++ b/codex-rs/core/templates/parallel/instructions.md @@ -0,0 +1,14 @@ +## Exploration and reading files + +- **Think first.** Before any tool call, decide ALL files/resources you will need. +- **Batch everything.** If you need multiple files (even from different places), read them together. +- **multi_tool_use.parallel** Use `multi_tool_use.parallel` to parallelize tool calls and only this. +- **Only make sequential calls if you truly cannot know the next file without seeing a result first.** +- **Workflow:** (a) plan all needed reads → (b) issue one parallel batch → (c) analyze results → (d) repeat if new, unpredictable reads arise. + +**Additional notes**: +* Always maximize parallelism. Never read files one-by-one unless logically unavoidable. +* This concern every read/list/search operations including, but not only, `cat`, `rg`, `sed`, `ls`, `git show`, `nl`, `wc`, ... +* Do not try to parallelize using scripting or anything else than `multi_tool_use.parallel`. + +## Editing constraints \ No newline at end of file From e9e644a119f2d2e89e6eb45db07d3da27a30ce33 Mon Sep 17 00:00:00 2001 From: zhao-oai Date: Tue, 18 Nov 2025 09:28:26 -0800 Subject: [PATCH 10/10] fixing localshell tool calls (#6823) - Local-shell tool responses were always tagged as `ExecCommandSource::UserShell` because handler would call `run_exec_like` with `is_user_shell_cmd` set to true. - Treat `ToolPayload::LocalShell` the same as other model generated shell tool calls by deleting `is_user_shell_cmd` from `run_exec_like` (since actual user shell commands follow a separate code path) --- codex-rs/core/src/tools/handlers/shell.rs | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-) diff --git a/codex-rs/core/src/tools/handlers/shell.rs b/codex-rs/core/src/tools/handlers/shell.rs index 173a3ccffa..43b2bb1299 100644 --- a/codex-rs/core/src/tools/handlers/shell.rs +++ b/codex-rs/core/src/tools/handlers/shell.rs @@ -117,7 +117,6 @@ impl ToolHandler for ShellHandler { turn, tracker, call_id, - false, ) .await } @@ -130,7 +129,6 @@ impl ToolHandler for ShellHandler { turn, tracker, call_id, - true, ) .await } @@ -178,7 +176,6 @@ impl ToolHandler for ShellCommandHandler { turn, tracker, call_id, - false, ) .await } @@ -192,7 +189,6 @@ impl ShellHandler { turn: Arc, tracker: crate::tools::context::SharedTurnDiffTracker, call_id: String, - is_user_shell_command: bool, ) -> Result { // Approval policy guard for explicit escalation in non-OnRequest modes. if exec_params.with_escalated_permissions.unwrap_or(false) @@ -285,12 +281,7 @@ impl ShellHandler { } } - // Regular shell execution path. - let source = if is_user_shell_command { - ExecCommandSource::UserShell - } else { - ExecCommandSource::Agent - }; + let source = ExecCommandSource::Agent; let emitter = ToolEmitter::shell(exec_params.command.clone(), exec_params.cwd.clone(), source); let event_ctx = ToolEventCtx::new(session.as_ref(), turn.as_ref(), &call_id, None);