Reflect

Before following this workflow, read the Codex runtime contract. It defines plugin paths, model configuration, delegation, and persistence.

Reflect

Mine the current conversation for durable learnings, then route them into skill edits.

When to invoke

Invoke when the user says "reflect" or "$reflect". Skip when the conversation is trivial, off-topic, or already covered by an existing skill the parent followed correctly. One-offs are not learnings.

Process

1. Locate the active transcript

The parent finds its own transcript file before fanning out. Codex transcripts live under ~/.codex/sessions/YYYY/MM/DD/*.jsonl. Read only files whose first session_meta record has payload.cwd equal to the current workspace; never sweep unrelated workspaces. Do not glob across ~/.codex/sessions/. That crosses workspace boundaries and reads private chats from unrelated projects.

find ~/.codex/sessions -type f -name '*.jsonl' -print0 | xargs -0 ls -t | head -20

For each candidate, read the first JSONL line and require type == "session_meta" plus an exact payload.cwd match. Use payload.session_id to distinguish the current session from older candidates. If no path resolves, write a tight digest of the session and pass that instead.

2. Spawn three reviewers in parallel

One message, three spawn_agent calls using the default agent, explicit model and reasoning settings on each, and normal write-capable mode. Reviewers need MCP access for context lookups referenced in the transcript. Subagents inherit the parent's available MCP tools.

Pass each template verbatim, substituting the transcript path or digest where marked. Reviewers return findings in their final response.

3. Synthesize

One spawn_agent call using the default agent and configured reflect-judgment model (default gpt-6-astra @ high) in normal write-capable mode. The synthesizer spot-verifies citations and may need inherited MCP tools. Use references/synthesizer.md verbatim, with each reviewer's full output inlined where marked. The synthesizer returns a structured Accepted / Rejected / Backlog list.

4. Structural enforcement check

Sanity-check the synthesizer's Accepted list. For any item that would be enforced more reliably by a lint rule, script, metadata flag, or runtime check, move it from Accepted to Backlog. See the encode-lessons-in-structure principle skill.

5. Apply

Before applying any Accepted edit, present the synthesizer's full Accepted/Rejected/Backlog output to the user and wait for explicit approval. The user picks which subset to apply and may redirect routings. Skill changes affect every future agent in the org. Do not auto-apply.

Backlog items file to whatever devex / backlog tracker your team uses automatically. Only the Accepted list waits for approval.

For each approved Accepted item, follow the Routing field exactly:

• Trivial existing-skill edit (a one-line bullet, a tightened sentence, a stale fact corrected): parent does directly.
• Substantive existing-skill edit (a new section, a new pattern table, more than ~10 lines): hand to Codex's skill-creator skill skill and run its draft / test / iterate loop.
• tune description: <skill path> (the skill exists but did not trigger when it should have): hand to skill-creator and run its description-optimization loop.
• new skill via skill-creator: <kebab-name>: hand creation to skill-creator. Do not invent the shape ad hoc.

If your environment ships a SKILL.md validator, run it on every touched skill before declaring done. Skip this step if it doesn't.

6. Summarize for the user

Short list, no preamble:

• Edits applied: <skill path>. What changed, one line each.
• New skills created: <skill path>. One line each (rare).
• Backlog filed to the devex tracker: <issue title> (<tags>). One line each.
• Dropped: one line per rejected finding + reason from the synthesizer.