Skip to content

tobihagemann/turbo / claude/skills/peer-review

peer-review

Verified

Run an independent peer review via Codex. Use when the user asks to "peer review", "peer review my code", "peer review my plan", "get a second opinion", or "independent review".

FileKindSize
SKILL.mdskill md7.9 KB
references/subagent-wrapping.mdmarkdown1.8 KB

references/subagent-wrapping.md

View on GitHub
# Subagent Wrapping

If you are a subagent invoking `/peer-review`, follow these guardrails. They cover the subagent return contract; the codex run mechanics (timeout, force-background recovery, the `Monitor` ban) live in `/codex-exec`.

## Guardrails

1. **Wait synchronously, in this turn.** Do not return until codex's complete output is in hand, then forward it verbatim. If codex was force-backgrounded, `Read` its `-o` file repeatedly within this same turn until content appears there. On the fresh path `/codex-exec` requires, an empty file means the run is still going rather than that it returned nothing. You are not re-invoked by the completion `<task-notification>` after you return, so you cannot end your turn to wait for it. If you truly cannot finish in-turn, surface the task ID and cause verbatim and request the parent resume you via `SendMessage` once codex completes.
2. **Treat as peer-review failure** (not as a successful empty result): premature return before codex completes, `"Waiting for codex to finish"` text, empty output, partial output (missing dimensions in a multi-dimension fan-out), a background ID returned in place of findings, or off-topic output that addresses code, files, or topics outside the reviewed material instead of the requested findings.
3. **Report the failure cause specifically** (usage limit, timeout, force-background without recovery, error message) when applicable.

## Failure Reporting

`/peer-review` Step 4 specifies an `AskUserQuestion` gate for failure modes. You cannot reach the user. If codex returned partial output, emit Step 4's "Incomplete output" notice verbatim and name what is missing. If codex returned nothing, failed, or returned off-topic output, emit Step 4's "No output / codex failed" notice verbatim. Do not synthesize an empty no-finding peer review.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall