Generate reflections for past Codex session histories using the AutoSkill Reflection CLI. This skill helps surface repeated patterns, friction, and potential skill ideas from prior Codex conversations.
What’s included
SKILL.md: guidance on when and how to use the skillscripts/: the CLI tools (reflect_sessions.pyand helpers)references/: command catalog, usage notes, and examples
Requirements
- Python 3.11+
- A local Codex session archive in
~/.codex/sessions - The
codexbinary available on your PATH (or pass--codex-path)
Quick start
Run commands from scripts/:
python3 reflect_sessions.py --output -
Human-readable output:
python3 reflect_sessions.py --output-style human --output -
Project filter example:
python3 reflect_sessions.py --project ExampleProject --output -
Preset prompt example:
python3 reflect_sessions.py --prompt-preset summary --output -
Inline prompt example:
python3 reflect_sessions.py --prompt-text "Summarize in 5 bullets." --output -Prompt presets
Available presets:
reflection(default): full reflection on repetition, friction, and skill ideassummary: concise summary of goals, actions, outputs, and decisionsbloat: bloat/dead ends/cleanup opportunities introduced during the sessionincomplete: open loops and unfinished tasksdecisions: key decisions, alternatives, and rationalenext_steps: concrete follow-up actions, tests, and validations
Use --prompt-preset <name>, --prompt-text "<prompt>", or --prompt-file /path/to/prompt.md.
Cache behavior
Reflections are cached per session and prompt. Cache files live here:
~/.codex/sessions/reflection_cache/<session_id>-<prompt_key>.json
prompt_key is a short hash derived from the prompt label (preset path or
inline:<hash> for inline prompts). Legacy cache files without the prompt key
are still read for the default reflection preset.
Notes on privacy
The CLI reads local session histories from ~/.codex/sessions. Reflections may contain
sensitive content from those sessions. Review outputs before sharing them publicly.
License
MIT. See LICENSE.