Codex CLI Optimizer Skill
Rules and tracking for Codex CLI configuration optimization.
See frontmatter schema: ~/ghq/github.com/i9wa4/internal/docs/schema/frontmatter-schema.md
1. Config File Location
User's Codex CLI config is stored at: @~/ghq/github.com/i9wa4/dotfiles/config/agents/ @~/ghq/github.com/i9wa4/dotfiles/config/codex/
Key files:
| File | Git | Description |
| -------------------- | --- | ----------------------------------------- |
| config.common.toml | Yes | Common settings (model, MCP, etc.) |
| AGENTS.md | No | Symlink to ../claude/CLAUDE.md |
| skills/ | No | Symlink to ../claude/skills/ |
| config.toml | No | Manually managed (common + trusted repos) |
2. Config Management
2.1. Why separate common and full config?
config.toml contains full paths for trusted projects:
[projects."/Users/uma/ghq/github.com/foo/bar"]
trust_level = "trusted"
This path includes username, so config.toml cannot be shared across
machines with different usernames. Solution:
config.common.toml- Shared settings (Git-managed)config.toml- Manually managed, machine-specific (Git-ignored)
2.2. How to update
- Edit
config.common.tomlfor shared settings - Copy shared settings to
config.tomlmanually - Add project-specific
[projects."..."]entries as needed
3. Fetch Releases
Use gh command to fetch the latest releases:
FILE=$(mkoutput --dir tmp --label output)
gh api repos/openai/codex/releases --jq '.[0:5]' > "$FILE"
cat "$FILE"
4. CHANGELOG Operations
4.1. Latest Release Summary
- Fetch releases using the command above
- Extract the first release entry
- Categorize changes into:
- New features (Added)
- Bug fixes (Fixed)
- Improvements (Improved/Changed)
- Breaking changes
- Present in Japanese with brief explanations
4.2. Version Diff
- Ask user for start and end versions
- Extract all releases between those versions
- Summarize cumulative changes
- Highlight breaking changes and deprecations
4.3. Breaking Changes Detection
- Search for keywords:
breaking,removed,deprecated,changed(behavior changes) - List affected commands and options
- Provide migration guidance
5. Settings Categories
| Category | Examples |
| --------- | ---------------------------------------------------- |
| Model | model, model_reasoning, model_reasoning_effort |
| Behavior | approval_mode, sandbox, network_access |
| Display | notify, tui.notifications_method |
| Shell | shell_environment_commands |
| History | history, project_doc_max_bytes |
| Features | features.skills, features.web_search_request |
| Disable | disable_response_storage, hide_agent_* |
| Analytics | analytics.enabled, feedback.enabled |
6. AGENTS.md Design Guidelines
- YOU MUST: Focus only on persona and core guidelines
- YOU MUST: Keep instructions compatible with both Claude Code and Codex CLI
- NEVER: Include tool-specific syntax (e.g., Claude Code's TodoWrite)
7. Optimization Checklist
Check the following when editing AGENTS.md or config.toml:
- [ ] Is the persona definition concise?
- [ ] Are basic rules truly needed at all times?
- [ ] Can detailed explanations be moved to skills/?
- [ ] Is config.toml using appropriate approval_mode?
8. Optimization Tracking
Last reviewed Codex CLI version: v0.104.0 (2026-02-23)
8.1. Applied Optimizations
- [x] AGENTS.md symlinked to CLAUDE.md (shared persona)
- [x] skills/ symlinked to Claude Code skills
- [x] config.common.toml for shared settings
8.2. Pending Considerations
- [ ] Create prompts/ symlink to
../claude/commands/if needed - [ ] Create generate-config.sh for automated config.toml generation
- [ ] Evaluate MCP server integration (AWS docs MCP etc.)
- [ ] Review
tui.notifications_methodconfig option (added in v0.93.0) - [x] Plan mode (
/plan) - now stable and enabled by default (v0.94.0) - [ ] Review smart approvals default behavior (enabled in v0.93.0)
- [ ]
personalitysetting - now stable (default: "friendly"); consider setting explicitly or usingpersonality = "pragmatic"(v0.94.0) - [ ]
AfterToolUsehook - new hook event for post-tool automation (v0.99.0) - [x]
command_attribution = "disable"- co-author attribution disabled (v0.103.0) - [ ]
approval_policy: on-failure- deprecated; review if used (v0.102.0) - [ ]
log_dirconfig - redirect logs to custom directory (v0.97.0)
8.3. Version Notes
- v0.104.0:
WS_PROXY/WSS_PROXYenv for websocket proxying,ConfigChangehook from app-server, command approval ID plumbing - v0.103.0: Commit co-author attribution via
command_attributionconfig, richer app details in listings - v0.102.0: Unified permissions flow with TUI history, slash command to grant
sandbox read access, customizable multi-agent roles,
model/reroutednotification,approval_policy: on-failuredeprecated - v0.101.0: Bug fixes for memory and model resolution
- v0.100.0: Experimental JS REPL runtime, memory slash commands
(
/m_update,/m_drop),ReadOnlyAccesssandbox policy,AfterToolUsehook - v0.99.0:
/statuslinecommand for TUI footer config, shell commands no longer interrupt in-flight turns, GIF/WebP image support,AfterToolUsehook added, security fix (RUSTSEC-2026-0009) - v0.98.0: GPT-5.3-Codex model introduced, steer mode stable by default (Enter sends, Tab queues)
- v0.97.0: "Allow and remember" for MCP/App tools, live skill updates,
/debug-configslash command, configurablelog_dir - v0.96.0:
unified_execenabled on all non-Windows platforms - v0.95.0:
~/.agents/skillspersonal skill loading,/planaccepts args and images, parallel shell tools,CODEX_THREAD_IDenv injected - v0.94.0: Plan mode enabled by default,
personalityconfig stable (default: friendly), skills from.agents/skills
9. Response Format (CHANGELOG)
# Codex CLI vX.X.X 更新内容
## 新機能
- [機能名]: [説明]
## バグ修正
- [修正内容]
## 改善
- [改善内容]
## 破壊的変更
- [影響を受ける機能と移行方法]
---
Source: https://github.com/openai/codex/releases
10. Important Notes
- Codex CLI changelog is on the releases page, not CHANGELOG.md
- Use GitHub releases API to fetch release notes
- Always specify the version being referenced
- Link to official release page when available
11. Reference Links
- Codex CLI: https://github.com/openai/codex