shishiv Reflect
Mine the current conversation for durable learnings, then route them into skill edits.
When to invoke
- The user said "reflect" or "/skill:pstack-reflect".
- A complex task (5+ tool calls) just landed cleanly and the recipe is worth keeping.
- The agent hit dead ends, found the working path, and the path generalizes.
- The user corrected the agent's approach mid-task.
- A non-trivial workflow emerged that isn't captured anywhere.
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 before fanning out. Use only the active session path named by $PI_SESSION_FILE. Do not infer another storage layout or glob for nearby transcripts, because that crosses workspace boundaries and reads private chats from unrelated projects. If the path is unavailable, write a tight digest of the session and pass that instead.
2. Spawn three reviewers in parallel
Follow ../../docs/delegation.md and launch three read-only reviewers together. Give external-source tools only to reviewers that need them. The prompt forbids file writes; the parent applies edits. Run capability preflight before launch.
| Lens | Role | Prompt template |
|---|---|---|
| Judgment | judgment with optional external-source tools | references/judgment-reviewer.md |
| Tooling | review with optional external-source tools | references/tooling-reviewer.md |
| Divergent | review | references/divergent-reviewer.md |
Pass each template verbatim, substituting the transcript path or digest where marked. Reviewers return findings through the selected delegation capability.
3. Synthesize
Delegate one judgment pass after the three reviews. Give it available read-only external-source tools when citation verification needs them, and forbid file writes. 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. The synthesizer already applies this criterion; this is a final pass before edits land. 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. Those are tracker submissions, not skill edits. 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 the host's skill-authoring workflow and run its draft / test / iterate loop.
tune description: <skill path>(the skill exists but didn't trigger when it should have): hand tocreate-skilland run its description-optimization loop.new skill via create-skill: <kebab-name>: hand creation tocreate-skill. 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.