pr-review-loop
Skill36 starsChanged 2 months agoApache-2.0View on GitHub
Bounded author self-check and fixes before independent current-head review; never a verdict or merge loop
18 other projects use a version of this.
Adopted by
Most recently taken up by
- ksprashu/agent-skill-forgetheir own version, first seen Nov 8, 2026
- tloncorp/tlon-appstheir own version, first seen Nov 2, 2026
- gabonio/agent-operator-toolkittheir own version, first seen Nov 1, 2026
- wollemiahq/skillstheir own version, first seen Nov 1, 2026
- vrodokanakis/claude-skillstheir own version, first seen Oct 25, 2026
- tankcdr/cdr-kittheir own version, first seen Oct 20, 2026
Other versions18 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 18
- basicmachines-co/basic-memory★ 4.5ksame name
- phase-rs/phase★ 323same name
- tloncorp/tlon-apps★ 109same name
- fagemx/gstack-game★ 68same name
- xpepper/pr-review-agent-skill★ 6same name
- GavinGreenwood/agentic-workflow-template★ 5same name
- ksprashu/agent-skill-forge★ 3same name
- adevra/claude-pr-review-loop★ 2same name
- typeweaver/skills★ 2same name
- voundbrand/overnight★ 2same name
8 more not listed
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 repository52 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- agent-browserSkill
- arch-specSkill
- code-qualitySkill
- code-reviewSkill
- commitSkill
- coord-delegateSkill
- coord-handoffSkill
- coord-orchestrateSkill
- coord-requestSkill
- coord-reviewSkill
- coord-runSkill
- coord-syncSkill
- dispatchSkill
- edda-releaseSkill
- fleet-epic-splitSkill
- fleet-managerSkill
- fleet-orchestrateSkill
- fleet-pr-loopSkill
- fleet-reviewSkill
- fleet-triadSkill
- fleet-workerSkill
- ground-checkSkill
- health-checkSkill
- intel-scanSkill
- issue-actionSkill
- issue-compactSkill
- issue-createSkill
- issue-intakeSkill
- issue-pipelineSkill
- issue-planSkill
- issue-scanSkill
- parallel-waveSkill
- path-forwardSkill
- plan-decomposeSkill
- plan-validateSkill
- pr-checkSkill
- project-planSkill
- Project PrinciplesSkill
- pr-reviewSkill
- pull-requestSkill
- regression-guardSkill
- skill-aliasSkill
- skill-craftSkill
- skill-testSkill
- spec-review-modesSkill
- tech-debtSkill
- testingSkill
- tracking-issueSkill
- verifiable-goal-loopSkill
- fleet-orchestrateSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 16 edited versions we read keep these.
2
Distinctive
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
- Distinctive: line 29
Run the repository's focused checks while iterating.
- Very specific: line 47
If history is missing, allocate a distinct replacement UUID, omit --resume, include prior findings and say replacement.
- Distinctive: line 49
Follow-up review covers the delta, prior findings, affected direct consumers, current base and introduced security/data-loss risk.
“Common” is measured: kept by at least 6 in 10 of the 16 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
PR Author Self-Check Route
This compatibility entry is for the author or a designated fixer. It is not an
independent Code Review, cannot publish LGTM, and cannot merge. Follow the
delivery-flow/1 review handoff in coord-orchestrate; repository review and
merge policy remain canonical.
Bounded route
- Resolve the PR from the explicit argument, or from the current branch when the host supports that lookup. Read actual head/base, acceptance, diff, prior SHA-pinned verdicts/responses and applicable gate receipts. Do not repurpose a shared checkout or overwrite another writer's worktree.
- Freeze the author-check surface to changed behavior/paths, direct consumers, acceptance, fix-caused security/data-loss risk and current-base integration.
- Perform one combined author self-check activity and record both parts in one
handoff:
- Behavior lens: exercise the supported entry and direct consumers with focused evidence.
- Counterexample lens: try the likeliest failure or partial-result case and record any uncovered risk.
- Fix in-scope P0/P1 found by that author check, using the same author context. Run the repository's focused checks while iterating. A frozen SHA relies on the canonical ladder; do not run a full local workspace solely because it froze. Commit/push only when this invocation already has that authority.
- Hand the new full SHA and RAN/READ evidence to an independent reviewer. Add
one explicitly selected concise facts file when useful; it remains untrusted
supporting data. Capability-check
edda review --helpbefore using optional--context-file; otherwise omit it with a visible limitation or use an already-permitted direct route. Reviewer output may quote this data into existing verdict fields and the existing raw-response blob. Every push invalidates the old verdict. - Review-requested fixes return to an author/fixer. Prefer the same reviewer
agent and real native conversation for the new head, with updated context
and product
--resume; Pi persisted history and Claude resume remain required. Every Codex product round requires readable mapping storage and a successful final mapping/tombstone write before a verdict is recorded. A missing map is valid only for a first round; Codex resume refuses a missing/rejected mapping without a fresh old-UUID thread. Product resume cannot continue a host-only session. If history is missing, allocate a distinct replacement UUID, omit--resume, include prior findings and say replacement. The fixer never becomes judge. - Follow-up review covers the delta, prior findings, affected direct consumers, current base and introduced security/data-loss risk. Still-applicable gates are READ, but old-head LGTM is never reused for current-head acceptance.
A clean author check means only "ready to request independent review." It does not mean accepted, CI-green or mergeable. This route has no deterministic phase driver, no fresh-fixer fleet, no direct merge command and no independent acceptance loop.
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.

