Persist Guardian V2 risk scores without restoring them (#40884)

## What changed

- Append accepted Guardian V2 classification results to rollout history for non-ephemeral threads.
- Keep resumed and forked threads from restoring a persisted score into active Guardian state.

## Testing

- Verify that asynchronous scoring records the resulting `SecurityRiskScore` rollout item.
- Seed resume and fork tests with a persisted score and verify that approvals ignore it.

GitOrigin-RevId: 40d63a2ea7e7cf6398402c6aed7d5e5727dc9d68

codex-rs/Cargo.lock

@@ -3329,6 +3329,7 @@ dependencies = [
  "codex-core",
  "codex-extension-api",
  "codex-features",
+ "codex-history",
  "codex-http-client",
  "codex-login",
  "codex-model-provider",

codex-rs/app-server/tests/suite/v2/guardian_v2.rs

@@ -1,3 +1,4 @@
+use std::io::Write;
 use std::sync::Arc;
 use std::sync::Mutex;
 use std::sync::atomic::AtomicUsize;
@@ -8,6 +9,7 @@ use anyhow::Result;
 use app_test_support::MockResponsesConfig;
 use app_test_support::TestAppServer;
 use app_test_support::create_fake_rollout;
+use app_test_support::rollout_path;
 use axum::Json;
 use axum::Router;
 use axum::extract::State;
@@ -571,14 +573,34 @@ async fn guardian_v2_routes_scoped_tool_approvals(
         | ThreadLifecycle::RootUserRestriction
         | ThreadLifecycle::RootUserInputRestriction
         | ThreadLifecycle::RootUserInputHookBlocked => None,
-        ThreadLifecycle::Resume | ThreadLifecycle::Fork => Some(create_fake_rollout(
-            codex_home.path(),
-            "2025-01-05T12-00-00",
-            "2025-01-05T12:00:00Z",
-            USER_CONTEXT,
-            Some("mock_provider"),
-            /*git_info*/ None,
-        )?),
+        ThreadLifecycle::Resume | ThreadLifecycle::Fork => {
+            let thread_id = create_fake_rollout(
+                codex_home.path(),
+                "2025-01-05T12-00-00",
+                "2025-01-05T12:00:00Z",
+                USER_CONTEXT,
+                Some("mock_provider"),
+                /*git_info*/ None,
+            )?;
+            let mut rollout = std::fs::OpenOptions::new().append(true).open(rollout_path(
+                codex_home.path(),
+                "2025-01-05T12-00-00",
+                &thread_id,
+            ))?;
+            writeln!(
+                rollout,
+                "{}",
+                json!({
+                    "timestamp": "2025-01-05T12:00:00Z",
+                    "type": "security_risk_score",
+                    "payload": {
+                        "scores": { "action_risk": 0.0 },
+                        "sampled_at": "2025-01-05T12:00:00Z",
+                    },
+                })
+            )?;
+            Some(thread_id)
+        }
     };
     let mut app_server = TestAppServer::builder()
         .with_codex_home(codex_home.path())
@@ -1143,7 +1165,7 @@ async fn guardian_v2_required_model_bypasses_scoring_and_runs_full_reviews() ->
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn resumed_thread_starts_without_guardian_score() -> Result<()> {
+async fn resumed_thread_ignores_persisted_guardian_score() -> Result<()> {
     skip_if_no_network!(Ok(()));
     guardian_v2_routes_tool_approvals(
         GuardianRisk::Low,
@@ -1156,7 +1178,7 @@ async fn resumed_thread_starts_without_guardian_score() -> Result<()> {
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn forked_thread_starts_without_guardian_score() -> Result<()> {
+async fn forked_thread_ignores_persisted_guardian_score() -> Result<()> {
     skip_if_no_network!(Ok(()));
     guardian_v2_routes_tool_approvals(
         GuardianRisk::Low,

codex-rs/core/src/session/session.rs

@@ -31,7 +31,6 @@ use codex_protocol::protocol::MultiAgentVersion;
 use codex_protocol::protocol::ThreadHistoryMode;
 use codex_protocol::protocol::ThreadSource;
 use codex_protocol::protocol::TurnEnvironmentSelections;
-use codex_protocol::security_risk::SecurityRiskScore;
 use codex_skills::SkillError;
 use std::sync::OnceLock;
 use tokio::sync::Semaphore;
@@ -752,18 +751,6 @@ impl Session {
             config.current_time_reminder.as_ref(),
             external_time_provider,
         )?;
-        if thread_extension_init.get::<SecurityRiskScore>().is_none()
-            && let Some(score) = initial_history
-                .get_rollout_items()
-                .iter()
-                .rev()
-                .find_map(|item| match item {
-                    RolloutItem::SecurityRiskScore(score) => Some(score),
-                    _ => None,
-                })
-        {
-            thread_extension_init.insert(score.clone());
-        }
         let selected_capability_roots =
             match thread_extension_init.get::<Vec<SelectedCapabilityRoot>>() {
                 Some(roots) => roots.as_ref().clone(),

codex-rs/ext/guardian-v2/Cargo.toml

@@ -17,6 +17,7 @@ codex-api = { workspace = true }
 codex-core = { workspace = true }
 codex-extension-api = { workspace = true }
 codex-features = { workspace = true }
+codex-history = { workspace = true }
 codex-http-client = { workspace = true }
 codex-login = { workspace = true }
 codex-model-provider = { workspace = true }

codex-rs/ext/guardian-v2/src/async_scorer/extension.rs

@@ -29,6 +29,7 @@ use codex_extension_api::ToolName;
 use codex_extension_api::ToolPayload;
 use codex_extension_api::ToolStartInput;
 use codex_features::Feature;
+use codex_history::RolloutItem;
 use codex_login::AgentIdentityAuthPolicy;
 use codex_login::AuthManager;
 use codex_model_provider::create_model_provider;
@@ -796,6 +797,19 @@ impl GuardianV2Extension {
                     .latest_scored_tool_call
                     .fetch_max(tool_call_index, Ordering::Release);
                 classification_finished_at = Some(Instant::now());
+                if !config.ephemeral
+                    && let Err(error) = thread
+                        .append_rollout_items(&[RolloutItem::SecurityRiskScore(score)])
+                        .await
+                {
+                    tracing::warn!(
+                        %thread_id,
+                        %turn_id,
+                        %call_id,
+                        %error,
+                        "failed to persist Guardian V2 classification result"
+                    );
+                }
                 Ok("success")
             }
             .await;

codex-rs/ext/guardian-v2/src/async_scorer/extension_tests.rs

@@ -22,6 +22,7 @@ use codex_extension_api::ToolName;
 use codex_extension_api::ToolPayload;
 use codex_extension_api::ToolStartInput;
 use codex_features::Feature;
+use codex_history::RolloutItem;
 use codex_login::AuthManager;
 use codex_login::CodexAuth;
 use codex_login::ExternalAuth;
@@ -1726,6 +1727,27 @@ async fn contributor_samples_tool_calls_with_the_existing_luna_pool() -> Result<
         }
     );
     assert!(score.sampled_at.is_some());
+    test.codex.ensure_rollout_materialized().await;
+    let persisted_score = tokio::time::timeout(Duration::from_secs(5), async {
+        loop {
+            if let Some(persisted_score) = test
+                .codex
+                .load_history(/*include_archived*/ false)
+                .await?
+                .items
+                .into_iter()
+                .find_map(|item| match item {
+                    RolloutItem::SecurityRiskScore(score) => Some(score),
+                    _ => None,
+                })
+            {
+                return Ok::<_, anyhow::Error>(persisted_score);
+            }
+            tokio::task::yield_now().await;
+        }
+    })
+    .await??;
+    assert_eq!(&persisted_score, score.as_ref());
     assert_eq!(
         registry
             .fast_approval_decision(