ce-code-review
Review a named diff or PR for bugs, regressions, tests, and standards. Use when asked to review code or when a shipping skill needs a review receipt. Use when asked to apply this review's findings locally. Use ce-resolve-pr-feedback for feedback already left on a PR.
- License 许可
- MIT
- 最近更新
- 2026-11-14
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- ce-code-review
- description
- Review a named diff or PR for bugs, regressions, tests, and standards. Use when asked to review code or when a shipping skill needs a review receipt. Use when asked to apply this review's findings locally. Use ce-resolve-pr-feedback for feedback already left on a PR.
- argument-hint
- [mode:agent] [apply:local] [blank to review current branch, or provide PR link]
02 资源文件 38
- 参考文档references/action-class-rubric.md6.5 KB
- 参考文档references/cross-model-recovery.md3.5 KB
- 参考文档references/cross-model-review.md7.4 KB
- 参考文档references/depth-paths.md8.5 KB
- 参考文档references/diff-scope.md5.8 KB
- 参考文档references/dispatch-reviewers.md27.4 KB
- 参考文档references/findings-schema.json7.8 KB
- 参考文档references/finish-input.md19.7 KB
- 参考文档references/finish-merge.md17.7 KB
- 参考文档references/finish-report.md29.6 KB
- 参考文档references/intent-and-plan.md8.4 KB
- 参考文档references/modes-and-output.md18.0 KB
- 参考文档references/persona-catalog.md7.8 KB
- 参考文档references/personas/adversarial-reviewer.md10.6 KB
- 参考文档references/personas/agent-native-reviewer.md3.2 KB
- 参考文档references/personas/api-contract-reviewer.md4.2 KB
- 参考文档references/personas/correctness-reviewer.md7.9 KB
- 参考文档references/personas/data-migration-reviewer.md5.2 KB
- 参考文档references/personas/deployment-verification-agent.md4.9 KB
- 参考文档references/personas/frontend-races-reviewer.md3.4 KB
- 参考文档references/personas/learnings-researcher.md18.3 KB
- 参考文档references/personas/maintainability-reviewer.md8.6 KB
- 参考文档references/personas/performance-reviewer.md3.5 KB
- 参考文档references/personas/previous-comments-reviewer.md3.5 KB
- 参考文档references/personas/project-standards-reviewer.md7.8 KB
- 参考文档references/personas/reliability-reviewer.md5.0 KB
- 参考文档references/personas/security-reviewer.md5.3 KB
- 参考文档references/personas/testing-reviewer.md7.0 KB
- 参考文档references/review-output-template.md13.3 KB
- 参考文档references/scope.md20.4 KB
- 参考文档references/select-and-route.md12.6 KB
- 参考文档references/subagent-model.md4.5 KB
- 参考文档references/subagent-template.md24.9 KB
- 参考文档references/validator-batch-template.md7.2 KB
- 脚本scripts/findings-mechanics.py9.8 KB
- 脚本scripts/packs-resolve.py32.0 KB
- 脚本scripts/review-scope.py15.6 KB
- 脚本scripts/run-log.py10.0 KB
03 SKILL.md 正文
Code Review
Help the caller deliver a correct change within the agreed scope. Find defects and improvements whose consequences justify action; judge the code against its intended behavior and project requirements, not a preferred rewrite. Serious defects remain important even when the diff is small. An adequate change needs no findings.
Done when: required review and validation are complete, retained findings are supported by the source, and the caller has a clear result with any remaining coverage limits. Apply only when the invocation authorizes it, under the rules below.
Artifact Root
<!-- ce-docs-root:start -->
Resolve the artifact root <root> before composing any artifact path.
- Read
docs_rootfrom<repo-root>/.rocketclaw/config.yamlonly (<repo-root>=jj workspace root). Do not read it fromconfig.local.yaml. Unset -><root>isdocs, exactly as before. - Validate a set value: a repo-relative directory whose real, symlink-resolved path stays inside the repo and is neither the repo root nor under
.git/or.jj/. Otherwise stop with an error namingdocs_rootand the value -- never fall back todocs. - Use
<root>as the sole artifact location: create it if absent, compose each path as<root>/<subdir>with this skill's own subdirectory, and never also readdocs. <!-- ce-docs-root:end -->
Execution spine
Follow these steps in order; the references supply the detail but never change the order. Read each reference when you enter the step whose own work it governs; a read made before that step does not satisfy it, and a reference you only hand to a leaf is not one you read.
- Read
references/modes-and-output.mdfirst. It settles arguments, conflicts, the quick-review short-circuit, the Review depth gate, and what this invocation returns. - Stage 1. Read
references/scope.mdand resolve the reviewed diff, the scope mode, and the deterministic scope signals. Then apply that Review depth gate before Stage 2. Lite ends the run without the later spine references; lite and focused each run from this context byreferences/depth-paths.md. - Stage 2. Read
references/intent-and-plan.md, write the intent summary every reviewer receives, and discover the plan Stage 6 verifies requirements against. - Stage 3. Read
references/persona-catalog.mdandreferences/select-and-route.md, then select the reviewers the change's risks call for, find the applicable standards files, and decide how the adversarial review will run. - Stage 3d. Read
references/subagent-model.mdand, when adversarial is selected locally,references/cross-model-review.md; resolve the native peer before local dispatch. This skill explicitly requests reviewers, satisfying a user-OR-skill rule without redundant approval, but never overriding an unconditional prohibition. Model selection and nesting depth are separate: omit forbidden model arguments and use a suitable inherited reviewer; accurately classify actual depth/denial/model/capacity/tool errors. Required unavailable cross-model independence or a fixed route is incomplete, not fully verified. Honor egress restrictions, explicit opt-outs andcross_model_review_mode: offwithout live opt-in. Disclose the allowed recipient/code transfer without another confirmation; collect terminal evidence before claiming coverage. Avoid duplicate peer/local adversarial reads and apply bounded native recovery only as the references permit. - Stage 4. Read
references/dispatch-reviewers.md. Dispatch the selected native reviewers concurrently within the host's cap and collect every launch's terminal result within this turn using bounded native collection. Launch acknowledgements and progress are not results. Record malformed returns and tool failures; stop/release uncollectable owned work, including peers, before returning failure. No detached CLI dispatch, polling or scheduled wakeups. Native peer collection may overlap this batch. - Stages 5 and 6. After collection write
references/finish-input.md's disk handoff. Normally dispatch its merge leaf, collect the independent validator batch it selects, then dispatch the report leaf; leaves launch no subagents and the report returns verbatim. Apply the same permission/model/depth gates to each launch. When permitted parent merge/report fallback is necessary, read the owned finish reference and use the same handoff/mechanical contracts, never inventing independent review or validation. Unavailable required validators remain degraded gates and required cross-model coverage remains incomplete.
Operating principles
- Report-only by default; never push. A bare
ce-code-reviewinvocation produces findings and does not apply them. Entering the apply stage requiresapply:local, or an explicit user request in the invoking prompt to apply or fix this review's findings; a deprecatedmode:autofixtoken is neither.mode:agentnever mutates the tree, even when nested inside a workflow that later applies findings. Never push, open PRs, or file tickets in any mode. - No blocking prompts. Never use
AskUserQuestion,request_user_input,ask_user, or other blocking question tools. Infer intent, plan, and scope from explicit tokens, git state, PR metadata, and conversation. Note uncertainty in Coverage or the verdict — do not stop to ask. - A spawned worker or peer starts a new context and does not continue the parent conversation.
- Explicit mutations only. Never run
gh pr checkout,git checkout,git switch, or similar branch-switch commands. Passing a PR number, URL, or branch name selects review scope, not permission to mutate the working tree. Uncommitted work can only be reviewed from the checkout that holds it, so to review it on a feature branch, stay on that branch (or check it out yourself) and passbase:or no target. - Report outcomes, not machinery. What you show the user is about the review: what is being examined, which coverage is included and the one-line reason for each conditional lens, the independent cross-model pass, and the findings. Name what the user would recognize, such as a PR number, a reviewer's concern, or a peer model. This skill's internal labels, dispatch bookkeeping, and setup narration stay out of user-facing text. Never claim more about the peer than its receipt attests. This governs what you surface and suppress, not the wording; use your own voice.
04 同仓库更多 Skills
ce-retro
Look back over an agent session in this repo, the current one by default or any scope the user names across harnesses, and find what in the repo's environment keeps costing agents effort or leading them into mistakes, keeping only fixes the evidence shows are worth it. Lands the approved environment fixes and hands the rest to their owners. Use for a retro, postmortem, or papercut sweep of how agents worked in this repo. Use ce-compound to keep the lesson of one solved problem, and ce-debug for a bug in product code.
release-notes
Draft release notes from merged pull requests since the last tag. Use when preparing a release.
skill-one
Sample skill
disabled-skill
A skill with model invocation disabled
claude-only-skill
A skill restricted to Claude Code only
agent-only-skill
An agent-only skill not intended for direct user invocation