Spawn three parallel review subagents over the active transcript, surface learnings, and route each to a concrete edit on an existing skill. Use when the user says reflect.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
On Codex, read the platform mapping, including its per-skill notes, before following this skill.
On GitHub Copilot, read the platform mapping, including its per-skill notes, before following this skill.
Mine the current conversation for durable learnings, then route them into skill edits.
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.
Before fanning out, find this conversation's transcript in Claude Code's per-project transcripts directory at ~/.claude/projects/<encoded-cwd>/. Do not glob across ~/.claude/projects/. That crosses workspace boundaries and reads private chats from unrelated projects.
Run the finder at skills/reflect/scripts/find-transcript.mjs under the installed plugin with the projects directory and a fragment of the conversation's opening user prompt:
node <plugin>/skills/reflect/scripts/find-transcript.mjs ~/.claude/projects/<encoded-cwd> "<opening prompt fragment>"
It covers the three layouts (flat <id>.jsonl, nested <id>/<id>.jsonl, subagent <parent>/subagents/<child>.jsonl), newest first, and prints the first path whose opening typed prompt carries the fragment. Do not reimplement the scan by hand: the first line of a transcript is session metadata, not a message, a session that starts with /clear or a ! shell command records that command's wrapper and output as user records before the prompt, and files run to several megabytes, so the finder streams each candidate and stops at its first typed user record. If it exits 1, write a tight digest of the session and pass that instead. It lists, one line each in turn order, the user's prompts and corrections, the tool calls and their results, and the files touched. It states no diagnosis, verdict, or cause. The reviewers judge the session, not your conclusions about it.
One message, three Agent calls, subagent_type: "general-purpose". Reviewers need MCP access to look up the tickets, chat threads, and observability traces the transcript references, so pick a subagent_type that retains MCP access. The prompt forbids file writes. The parent applies edits.
Each reviewer and the synthesizer name a role line in pstack-models.md and a default in Models. Set model to that line's value, or to the default if the sheet or the line is missing. Leave model unset when the value is auto or inherit-parent. If the Agent tool rejects a slug, use the default and say so. If it rejects the default, use the closest valid slug of the same family from its error message.
| 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, substituting the transcript path or digest where marked. Add nothing beside a transcript path, not even a focus note for a long transcript. Reviewers return findings in the Agent response body.
One Agent call, subagent_type: "general-purpose", model from the reflect judgment, divergent, synthesizer line (default in Models). It spot-verifies citations through MCP, so pick a subagent_type that retains MCP access. Pass references/synthesizer.md verbatim, with each reviewer's full output inlined where marked. It returns an Accepted / Rejected / Backlog list.
Move any Accepted item that a lint rule, script, metadata flag, or runtime check would enforce more reliably to Backlog. See the encode-lessons-in-structure principle skill.
Present the synthesizer's full Accepted / Rejected / Backlog output and wait for explicit approval before applying any Accepted edit. The user picks the subset and may redirect routings. Skill changes affect every future agent in the org. Do not auto-apply.
File each Backlog item to your team's devex or backlog tracker without waiting. Only the Accepted list waits for approval.
Follow each approved row's Routing exactly:
tune description: <skill path> (the skill exists but didn't trigger when it should have): hand to plugin-dev:skill-development and run its description-optimization loop.new skill via plugin-dev:skill-development: <kebab-name>: hand creation to plugin-dev:skill-development. Do not invent the shape ad hoc.If your environment ships a SKILL.md validator, run it on every touched skill before declaring done.
Short list, no preamble:
<skill path>. What changed, one line each.<skill path>. One line each (rare).<issue title> (<tags>). One line each.Role defaults, stamped from plugins/pstack/models.json (edit there, rerun tools/generate.mjs). A matching role line in the pstack-models.md override sheet overrides each at runtime; /setup-pstack writes it and lists its path per runtime.
opusopusA role value in the override sheet may name a reasoning effort after its model, as in opus @xhigh. Levels on Claude Code: low, medium, high, xhigh, max. Which ones apply depends on the model. A value without @ takes the sheet's default effort line, a level or session, and session when the sheet has no such line. session sets no effort, so the dispatch is the usual one. Strip the suffix before reading the model: inherit-parent or auto still omits model at every level, and a model name is passed as model. On Claude Code, a level picks the effort agent from the subagent_type you would otherwise use. pstack:poteto-agent becomes subagent_type: "pstack:poteto-agent-<level>". general-purpose, or no subagent_type, becomes subagent_type: "pstack:effort-<level>". The effort agents set only effort, so the model you pass still decides the model. On Codex, pass the level as spawn_agent's reasoning_effort and keep the usual instructions.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in. Use for /architect, 'architect this', 'design this', or non-trivial work where jumping to code would lock in the wrong shape.
日本語の概要は準備中です。原文の説明を表示しています。
Spawn N parallel candidates at the same task, pick a base, graft the strongest parts of the losers into it. Use for /arena, 'arena this', 'throw it in the arena', or when one attempt at a non-trivial artifact would lock in the wrong shape.
日本語の概要は準備中です。原文の説明を表示しています。
Use for "automate me", "create/update/refresh my -mode skill", "turn/capture my preferences or working style into a skill", or wanting agents to follow how the user works. Drafts or revises a personal -mode skill via plugin-dev:skill-development + unslop, optionally pulling fresh evidence from recent transcripts.
日本語の概要は準備中です。原文の説明を表示しています。
Watch an open PR — fix failing CI, handle the straightforward review comments, and drive it to a mergeable state. Claude Code analog of Cursor's built-in /babysit. Use after opening a PR when the user wants the agent to shepherd it without re-prompting.
日本語の概要は準備中です。原文の説明を表示しています。
Vet a perf measurement (limiter, tuning, limits, errors, repeatability, relevance, and whether the work happened) before you report or act on it. Use when you run a benchmark or report a speedup or regression you measured.
日本語の概要は準備中です。原文の説明を表示しています。
Find what a change could break somewhere else before it ships, beyond the diff, and prove the one fact it's safe because of by running real code instead of writing it up. Use for 'blast radius of X', 'what could this break', reviewing a small diff you don't trust, or a brief that asserts something about existing code ('make X public', 'X already handles Y') before you design against that assertion.
日本語の概要は準備中です。原文の説明を表示しています。