Unique-Divine/jiyuu / ai-skills/gh-rev
gh-rev
VerifiedUse the gh-rev CLI to maintain a persistent Markdown review ledger for local branches and pull requests. Trigger when the user names gh-rev, asks to save or submit a durable review, inspect or implement existing findings, continue an independent review pass, or adopt branch history into a PR. Use the gh-pr-review skill for review content and judgment. Do not use this skill for a one-off review without a durable ledger.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.0 KB |
| references/cli.md | markdown | 3.8 KB |
| references/implement-review-feedback.md | markdown | 6.3 KB |
| references/review-local-branch.md | markdown | 2.4 KB |
| references/review-pull-request.md | markdown | 2.8 KB |
| references/work-revisions.md | markdown | 3.0 KB |
references/work-revisions.md
View on GitHub# Read and work existing revisions
Use this workflow when the goal is to understand review status, implement
requested changes, or resolve findings. This is a primary `gh-rev` use case,
not an afterthought to review creation.
## Example: report review status
For "Are there reviews for this PR?" or "What remains unresolved?":
1. Resolve the target with `gh-rev status`, using `--pr` for published PR state
or `--branch` for local branch state. Unlike `open`, `status` does not create
a pending review snapshot.
2. Read `context.md` and every `rev-*.md` beneath the returned `target_path`,
not only the newest file.
3. Query exact unresolved markers:
```bash
gh-rev status todo owner__repo --pr 1153
# or: gh-rev status todo owner__repo --branch feature/cache
```
4. Summarize review count, recommendations, and unresolved findings. Do not
infer approval merely from an empty current TODO list; resolved findings and
each revision's recommendation remain part of the history.
## Example: implement requested changes
For "Take the reviews, fix the requested changes, and document the result":
1. Read repository instructions, the issue or PR intent, `context.md`, every
revision, and `status todo`. Group duplicate or dependent findings before
editing.
2. Work in a local checkout whose branch and SHA correspond to the selected
target. If starting from an online-only PR, use the repository's normal
checkout/worktree flow to materialize it before editing.
3. Inspect the cited code and verify each finding rather than accepting it
mechanically. Implement the smallest coherent fix that satisfies the
underlying concern and repository conventions.
4. Run focused tests for every changed behavior, followed by the repository's
proportionate type, lint, build, or broader test checks.
5. Re-read the resulting diff and the originating finding. Only after evidence
shows it is handled, change its exact marker from `- [ ] rev:` to
`- [x] rev:` in the original `rev-N.md` and add an indented resolution note
naming the fix and verification.
6. If the finding is invalid or intentionally declined, resolve it with a note
containing the concrete evidence and decision. Resolution means the concern
has been considered and closed, not necessarily that its proposed edit was
accepted.
7. Rerun `gh-rev status todo` and report anything still unresolved.
Do not allocate a new revision just to document fixes to old findings. The
originating revision is the durable task record. Allocate a new revision only
when someone performs another independent review of the resulting code.
## Boundaries
- Editing source files requires the user's request to implement or change code;
a request to inspect or explain reviews is read-only.
- Checking a finding does not publish to GitHub or imply GitHub approval.
- Preserve review history. Never delete a finding because it was fixed,
disagreed with, superseded, or inconvenient.
- Use only unresolved and resolved states. Put nuance in the resolution note.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 472k | Updated | Claude CodeCodex | Verified | Install |
| 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: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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 available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |