e9fb49366c
- The TUI now has quick reasoning controls: `Alt+,` lowers reasoning, `Alt+.` raises it, and accepted model upgrades now reset reasoning to the new model’s default instead of carrying over stale settings. (#18866, #19085) - App-server sessions can now manage multiple environments and choose an environment and working directory per turn, which makes multi-workspace and remote setups easier to target precisely. (#18401, #18416) - Added first-class Amazon Bedrock support for OpenAI-compatible providers, including AWS SigV4 signing and AWS credential-based auth. (#17820) - Remote plugin marketplaces can now be listed and read directly, with more reliable detail lookups and larger result pages. (#18452, #19079) - Hooks are now stable, can be configured inline in `config.toml` and managed `requirements.toml`, and can observe MCP tools as well as `apply_patch` and long-running Bash sessions. (#18893, #18385, #18391, #18888, #19012) - Eligible ChatGPT plans now default to the Fast service tier unless you explicitly opt out. (#19053) ## Bug Fixes - Preserved Cloudflare cookies across approved ChatGPT hosts, reducing auth breakage in HTTP-backed ChatGPT flows. (#17783) - Fixed remote app-server reliability issues so websocket events keep draining under load and shutdown no longer fails when the remote worker exits during cleanup. (#18932, #18936) - Fixed permission-mode drift so `/permissions` changes survive side conversations and updated Full Access state is correctly reflected in MCP approval handling. (#18924, #19033) - Fixed `wait_agent` so it returns promptly when mailbox work is already queued instead of waiting for a fresh notification or timing out. (#18968) - Fixed local stdio MCP launches for relative commands without an explicit `cwd`, bringing fallback path resolution in line with CLI behavior. (#19031) - Startup now fails less often on managed config edge cases: unknown feature requirements warn instead of aborting, and cloud-requirements errors are clearer about what failed. (#19038, #19078) ## Changelog Full Changelog: https://github.com/openai/codex/compare/rust-v0.123.0...rust-v0.124.0 - #18870 Load app-server config through ConfigManager @pakrym-oai - #18866 feat(tui): shortcuts to change reasoning level temporarily @fcoury-oai - #18430 app-server: implement device key v2 methods @euroelessar - #18757 fix: fully revert agent identity runtime wiring @efrazer-oai - #17783 Preserve Cloudfare HTTP cookies in codex @shijie-oai - #18876 [rollout_trace] Add rollout trace crate @cassirer-openai - #18401 Support multiple managed environments @starr-openai - #18797 Allow guardian bare allow output @maja-openai - #18886 Normalize /statusline & /title items @canvrno-oai - #18768 [codex] Tighten external migration prompt tests @alexsong-oai - #18909 Update /statusline and /title snapshots @canvrno-oai - #18867 sandboxing: materialize cwd-relative permission globs @bolinfest - #18915 fix: windows snapshot for external_agent_config_migration::tests::prompt_snapshot did not match windows output @bolinfest - #18416 Add turn-scoped environment selections @starr-openai - #18391 fix(core): emit hooks for apply_patch edits @fcoury-oai - #18916 test(core): move prompt debug coverage to integration suite @bolinfest - #17820 feat: add AWS SigV4 auth for OpenAI-compatible model providers @celia-oai - #18913 bazel: run wrapped Rust unit test shards @bolinfest - #18452 feat: Support remote plugin list/read. @xl-openai - #18936 Fix remote app-server shutdown race @bolinfest - #18871 refactor: add agent identity crate @efrazer-oai - #18276 exec-server: carry filesystem sandbox profiles @bolinfest - #18926 ci: keep argument comment lint checks materialized @bolinfest - #18935 Keep TUI status surfaces in sync @etraut-openai - #18923 chore(tui) debug-config guardian_policy_config @dylan-hurd-oai - #18943 tests: serialize process-heavy Windows CI suites @bolinfest - #18934 [codex] Clean guardian instructions @dylan-hurd-oai - #18948 chore: remove unused Bedrock auth lazy loading @celia-oai - #18277 core: derive active permission profiles @bolinfest - #18785 feat: add explicit AgentIdentity auth mode @efrazer-oai - #18953 use long-lived sessions for codex sandbox windows @iceweasel-oai - #18278 app-server: expose thread permission profiles @bolinfest - #17693 [codex-analytics] guardian review analytics events emission @rhan-oai - #17695 [codex-analytics] guardian review truncation @rhan-oai - #17696 [codex-analytics] guardian review TTFT plumbing and emission @rhan-oai - #18962 nit: expose lib @jif-oai - #18502 Support multiple cwd filters for thread list @acrognale-oai - #18968 fix: wait_agent timeout for queued mailbox mail @jif-oai - #18971 fix: cargo deny @jif-oai - #18973 chore: prep memories for AB @jif-oai - #18852 [codex] Update imagegen system skill @vb-openai - #18865 Stage publishable Python runtime wheels @sdcoffey - #18932 TUI: Keep remote app-server events draining @etraut-openai - #18877 [rollout_trace] Record core session rollout traces @cassirer-openai - #18959 feat(auto-review) policy config @dylan-hurd-oai - #18955 Add plumbing to approve stored Auto-Review denials @won-openai - #18999 arg0: keep dispatch aliases alive during async main @bolinfest - #18925 feat: Fairly trim skill descriptions within context budget @xl-openai - #18890 feat(auto-review) short-circuit @dylan-hurd-oai - #18279 app-server: accept permission profile overrides @bolinfest - #18582 [2/4] Implement executor HTTP request runner @aibrahim-oai - #18197 feat: add guardian network approval trigger context @viyatb-oai - #19033 Fix MCP permission policy sync @leoshimo-oai - #19016 exec-server: expose arg0 alias root to fs sandbox @bolinfest - #19036 Overlay state DB git metadata for filtered thread lists @joeytrasatti-openai - #18956 [Codex] Register browser requirements feature keys @khoi-oai - #19043 Update bundled OpenAI Docs skill freshness check @kkahadze-oai - #18504 Rebrand approvals reviewer config to auto-review @won-openai - #19046 exec-server: require explicit filesystem sandbox cwd @bolinfest - #18280 clients: send permission profiles to app-server @bolinfest - #18281 rollout: persist turn permission profiles @bolinfest - #18888 hooks: emit Bash PostToolUse when exec_command completes via write_stdin @eternal-openai - #19056 Rename approvals reviewer variant to auto-review @won-openai - #18583 [3/4] Add executor-backed RMCP HTTP client @aibrahim-oai - #19059 core: box multi-agent wrapper futures @bolinfest - #19031 Fix relative stdio MCP cwd fallback @mzeng-openai - #19063 chore(auto-review) feature => stable @dylan-hurd-oai - #19050 feat(request-permissions) approve with strict review @dylan-hurd-oai - #19067 test: set Rust test thread stack size @bolinfest - #19072 tui: fix approvals popup disabled shortcut test @bolinfest - #18893 codex: support hooks in config.toml and requirements.toml @eternal-openai - #18282 protocol: report session permission profiles @bolinfest - #19053 Default Fast service tier for eligible ChatGPT plans @shijie-oai - #19055 Add safety check notification and error handling @etraut-openai - #18283 app-server: accept command permission profiles @bolinfest - #19012 Mark codex_hooks stable @abhinav-oai - #18924 TUI: preserve permission state after side conversations @etraut-openai - #19071 Add computer_use feature requirement key @leoshimo-oai - #19079 Use remote plugin IDs for detail reads and enlarge list pages @xl-openai - #19038 feat: Warn and continue on unknown feature requirements @xl-openai - #19078 Clarify cloud requirements error messages @gverma-openai - #19085 Persist target default reasoning on model upgrade @shijie-oai - #19086 app-server: include filesystem entries in permission requests @bolinfest - #18385 Support MCP tools in hooks @abhinav-oai - #19113 Fix auto-review config compatibility across protocol and SDK @won-openai
Codex CLI (Rust Implementation)
We provide Codex CLI as a standalone executable to ensure a zero-dependency install.
Installing Codex
Today, the easiest way to install Codex is via npm:
npm i -g @openai/codex
codex
You can also install via Homebrew (brew install --cask codex) or download a platform-specific release directly from our GitHub Releases.
Documentation quickstart
- First run with Codex? Start with
docs/getting-started.md(links to the walkthrough for prompts, keyboard shortcuts, and session management). - Want deeper control? See
docs/config.mdanddocs/install.md.
What's new in the Rust CLI
The Rust implementation is now the maintained Codex CLI and serves as the default experience. It includes a number of features that the legacy TypeScript CLI never supported.
Config
Codex supports a rich set of configuration options. Note that the Rust CLI uses config.toml instead of config.json. See docs/config.md for details.
Model Context Protocol Support
MCP client
Codex CLI functions as an MCP client that allows the Codex CLI and IDE extension to connect to MCP servers on startup. See the configuration documentation for details.
MCP server (experimental)
Codex can be launched as an MCP server by running codex mcp-server. This allows other MCP clients to use Codex as a tool for another agent.
Use the @modelcontextprotocol/inspector to try it out:
npx @modelcontextprotocol/inspector codex mcp-server
Use codex mcp to add/list/get/remove MCP server launchers defined in config.toml, and codex mcp-server to run the MCP server directly.
Notifications
You can enable notifications by configuring a script that is run whenever the agent finishes a turn. The notify documentation includes a detailed example that explains how to get desktop notifications via terminal-notifier on macOS. When Codex detects that it is running under WSL 2 inside Windows Terminal (WT_SESSION is set), the TUI automatically falls back to native Windows toast notifications so approval prompts and completed turns surface even though Windows Terminal does not implement OSC 9.
codex exec to run Codex programmatically/non-interactively
To run Codex non-interactively, run codex exec PROMPT (you can also pass the prompt via stdin) and Codex will work on your task until it decides that it is done and exits. If you provide both a prompt argument and piped stdin, Codex appends stdin as a <stdin> block after the prompt so patterns like echo "my output" | codex exec "Summarize this concisely" work naturally. Output is printed to the terminal directly. You can set the RUST_LOG environment variable to see more about what's going on.
Use codex exec --ephemeral ... to run without persisting session rollout files to disk.
Experimenting with the Codex Sandbox
To test to see what happens when a command is run under the sandbox provided by Codex, we provide the following subcommands in Codex CLI:
# macOS
codex sandbox macos [--full-auto] [--log-denials] [COMMAND]...
# Linux
codex sandbox linux [--full-auto] [COMMAND]...
# Windows
codex sandbox windows [--full-auto] [COMMAND]...
# Legacy aliases
codex debug seatbelt [--full-auto] [--log-denials] [COMMAND]...
codex debug landlock [--full-auto] [COMMAND]...
Selecting a sandbox policy via --sandbox
The Rust CLI exposes a dedicated --sandbox (-s) flag that lets you pick the sandbox policy without having to reach for the generic -c/--config option:
# Run Codex with the default, read-only sandbox
codex --sandbox read-only
# Allow the agent to write within the current workspace while still blocking network access
codex --sandbox workspace-write
# Danger! Disable sandboxing entirely (only do this if you are already running in a container or other isolated env)
codex --sandbox danger-full-access
The same setting can be persisted in ~/.codex/config.toml via the top-level sandbox_mode = "MODE" key, e.g. sandbox_mode = "workspace-write".
In workspace-write, Codex also includes ~/.codex/memories in its writable roots so memory maintenance does not require an extra approval.
Code Organization
This folder is the root of a Cargo workspace. It contains quite a bit of experimental code, but here are the key crates:
core/contains the business logic for Codex. Ultimately, we hope this to be a library crate that is generally useful for building other Rust/native applications that use Codex.exec/"headless" CLI for use in automation.tui/CLI that launches a fullscreen TUI built with Ratatui.cli/CLI multitool that provides the aforementioned CLIs via subcommands.
If you want to contribute or inspect behavior in detail, start by reading the module-level README.md files under each crate and run the project workspace from the top-level codex-rs directory so shared config, features, and build scripts stay aligned.