Self review

Skill ayaqen/Agents-Scripts/skills/self-review

Review your own diff like a hostile reviewer before committing or opening a PR.From its SKILL.md

Install

npx -y skills add ayaqen/Agents-Scripts --skill self-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

1.8 KB, as published. Nobody here has run it

self-review

A structured pass over your own changes before anyone (human or CI) sees them. Cheapest possible review round: the one that never happens because you caught it first.

When to use

Before every commit of non-trivial work.
Before marking a PR ready for review.
After an agent session produced a large diff quickly — speed of generation is not evidence of correctness.

Workflow

Read the full diff cold: git diff (or git diff --cached), top to bottom, as if someone else wrote it.
For each hunk ask, in order:
- Correctness: what input or state makes this line wrong? Check boundaries, empty cases, error paths, concurrency.
- Scope: does this hunk serve the stated task? Unrelated drive-by edits get split out or reverted.
- Leftovers: debug prints, commented-out code, TODOs without owners, stray files (git status for untracked strays).
- Consistency: naming and patterns match the surrounding file, not your habits.
Check the negative space — what the diff doesn't touch: callers of changed signatures, docs and changelog for user-visible behavior, tests for new branches, config for new options.
Re-run the narrowest proving command and read its output; don't rely on an earlier run from before your last edit.
Only findings fixed or consciously accepted? Commit with a message that says why, not just what.

Pitfalls

Reviewing from memory of what you intended instead of what the diff says. Read the actual bytes.
Rubber-stamping your own large diff because you're tired — that's exactly when step 2 matters. Take the hunks in reverse order to break the skim.
Fixing review findings and then not re-running the tests (step 4 applies again after every fix).