address-review-findings
Skill5 starsChanged 2 months agoApache-2.0View on GitHub
How a ship fix round addresses structured review findings — every severity, one disposition each, coherent history, truthful description, repush.
- Deletes or force-pushes
- Commits and pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- asarelcastellanos/claude-workbenchtheir own version, first seen Nov 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- asarelcastellanos/claude-workbench★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository11 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- code-review-and-qualitySkill
- doubt-driven-developmentSkill
- incremental-implementationSkill
- performance-optimizationSkill
- pr-descriptionSkill
- re-review-deltaSkill
- security-and-hardeningSkill
- spec-driven-developmentSkill
- test-driven-developmentSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
Follow the shared writing rules supplied by use_skill.
- Distinctive: line 25
Read every finding first and group by file — fixes in one file land together.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Addressing review findings (ship fix rounds)
You received structured findings from a review round (id, severity, file:line, title) plus the review's prose. Your job is to end the round with the PR merge-ready-or-honestly-disputed: every finding addressed or declined on the record, the history clean, the description true, the branch repushed.
The contract
Follow the shared writing rules supplied by use_skill. State the affected product task, the failure and the result of the fix. Load pr-description before updating the description; explain the product context for a reader who has never used the system.
- Address ALL severities, including
nit. A nit you agree with is a fix like any other. Only decline a finding you can argue against on the merits. - One disposition per finding, via
submit_dispositions.fixed— you changed the code (the note says what changed, one line).declined— you did not, and the note carries the actual argument (a reviewer reads it; "won't fix" is not an argument). Never skip a finding: an unaddressed finding counts against the pipeline's cap report, a declined one is an honest disagreement. - Fix at the root, not at the symptom. If a finding names an invariant and a case table, check every row (scenario → expected result) and add focused tests for the applicable paths. For a push-safety finding, inspect sources, destinations, endpoints and compound-command execution, not only the first example. Fix every instance of the pattern — the re-review reads the whole head, and a half-fixed pattern earns a widened finding under the same id.
- Squash to coherent commits. No "address review comments" trailer commits: fold each fix into the commit it amends (or one coherent fix-up commit when the fixes are cross-cutting). The history must read as if written right the first time.
- The description tells the truth. Resubmit through
submit_pr_descriptionwhen the change moved: Tour anchors that now point at shifted lines get corrected ranges (locate withgrep -n, verify against the file at your head — never from memory), new/changed behavior appears in the right sections, validation rows reflect what you actually re-ran. Switchboard re-renders and edits the PR body at your new head. - Repush, then submit. Push the squashed branch (
git push -fon the pipeline branch is expected after a squash), thensubmit_dispositions, thensubmit_pr_description. Run the tests before pushing; a fix round that repushes red wastes a full review round. - NEVER merge a pull request and NEVER approve one — no merge or approve command, no merge/approve API call, no pushing to the default branch. A human decides what merges.
Order of work
Read every finding first and group by file — fixes in one file land together. Fix, test, squash, push, dispositions, description, then your final summary: what you fixed, what you declined and why, in finding-id order.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

