Before resolving models or delegating, read T3 execution. It defines global configuration, capacity selection, and transport constraints.
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 transcript is the current T3 thread. Read it with t3_thread_read; use t3_thread_search to pull exact passages. Do not search across other threads. That crosses workspace boundaries and reads private chats from unrelated projects.
Reviewers get only their brief, never the parent's context. Pass the thread id and tell them to read it with t3_thread_read, or write a tight digest of the session and pass that instead.
2. Delegate three reviewers in parallel
One turn, three delegate_task calls with mode: "async". Reviewers need tool access for context lookups (tickets, chat threads, observability traces referenced in the transcript), so do not restrict their tools.
Resolve each model from the role line in pstack-models.md via orchestrator_capabilities. Never hardcode a slug.
| Lens | Role line | Prompt template |
|---|---|---|
| Judgment | reflect judgment, divergent, synthesizer |
references/judgment-reviewer.md |
| Tooling | reflect tooling |
references/tooling-reviewer.md |
| Divergent | reflect judgment, divergent, synthesizer |
references/divergent-reviewer.md |
Pass each template verbatim as the brief, substituting the thread id or digest where marked. Drain with task_status; reviewers return findings in their task result.
3. Synthesize
One delegate_task call, model from the reflect judgment, divergent, synthesizer line. The synthesizer's quality check includes spot-verifying citations, which can require tool access. Use references/synthesizer.md verbatim as the brief, 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): follow the authoring-a-skill playbook and run its draft / test / iterate loop.
tune description: <skill path>(the skill exists but didn't trigger when it should have): follow the authoring-a-skill playbook and run its description-optimization loop.new skill via authoring-a-skill: <kebab-name>: hand creation to the authoring-a-skill playbook. 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.