Code review
Tool-neutral AI Engineering System: 174 reusable skills (installable as Claude Code plugins) for planning and building software with AI agents. Classify → plan → approve → build under quality gates. Works with Claude Code, Codex, Cursor, Windsurf, Copilot & Antigravity. Canonical in .ai/.
npx -y skills add ahtishamshahzad/agent_dev_flow --skill code-reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 21 days oldThe repository was created 21 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 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.
What its author says it does
Copied from the file, not written here
Use to review a change or diff for correctness, clarity, maintainability, and convention-fit before it advances. Reports findings with severity, separating confirmed defects from suggestions; does not rubber-stamp.
SKILL.md
3.6 KB, 718 tokens by cl100k_base, as published. Nobody here has run it
Code Review
Purpose
Evaluate a change for correctness, readability, maintainability, and consistency with the project's conventions and architecture — and report actionable findings. Supports Gate 6.
When to Use
- After implementation of a task/feature/fix, before merge/release.
- For a code review request type.
- Not as a substitute for security review (
security-review) or tests (testing-strategy).
Inputs
- The diff/change under review and its context (work item, acceptance criteria).
- Project conventions and architecture.
Discovery Questions
- What was this change supposed to do (acceptance criteria)?
- Which files/modules changed, and what do they touch?
- Are there tests, and do they cover the change?
Responsibilities
- Verify the change meets its acceptance criteria.
- Check correctness (logic, edge cases, error handling).
- Check clarity/maintainability (naming, structure, duplication).
- Check convention/architecture fit and unintended behavior changes.
- Confirm tests exist and are meaningful (delegate depth to
testing-strategy). - Report findings with severity and confirmed vs suggestion.
Required Workflow
- Read acceptance criteria + the diff.
- Assess correctness and edge cases.
- Assess clarity, maintainability, conventions.
- Check tests cover the change.
- Flag security/perf concerns to the specialist skills.
- Report findings (severity, confirmed/suggestion) with
file:line.
Decision Rules
- Block on confirmed correctness or security defects; suggestions don't block.
- Prefer the smallest correct change; flag scope creep.
- Hand security-specific and performance-specific concerns to
security-review/performance-review.
Rules
- Be specific: cite
file:lineand the failure scenario. - Separate confirmed defects from style/preference.
- Don't approve unverified correctness; don't rubber-stamp.
Anti-Patterns
- Vague feedback with no location or scenario.
- Bikeshedding style while missing correctness.
- Approving without checking the change meets its criteria.
Validation Checklist
- Change meets acceptance criteria.
- Correctness/edge cases assessed.
- Clarity/maintainability assessed.
- Conventions/architecture respected.
- Tests cover the change.
- Findings have severity + confirmed/suggestion + location.
Definition of Done
A review report with located, severity-rated findings (confirmed defects vs suggestions), a clear pass/block verdict against acceptance criteria, and any security/perf items routed to specialists.
Related Skills
security-review, performance-review, testing-strategy, ai-output-review, final-quality-audit, project-orchestrator.
Related Knowledge
../../knowledge/ (conventions, architecture).
Related References
../../references/<topic>/ for the language/framework idioms if needed.
Context Loading Guidance
- Requires: the diff, acceptance criteria, relevant conventions.
- Does not require: the whole repo, unrelated references, planning skills.
- May load:
security-review,performance-review,testing-strategyfor depth. - Stop when: the review report is delivered.
Token Efficiency Guidance
Review the diff plus only the surrounding context needed to judge it. Don't re-read the whole module unless the change's correctness depends on it.
Gives 0 of the 12 instructions most code review skills give in 718 tokens
Counted across 610 of the 674 authors here whose files we hold, read 2026-08-06
- push back with technical reasoning if wrongin 60 of 610, across 24 files
- ask for clarification on unclear itemsin 51 of 610, across 16 files
- fix critical issues immediatelyin 45 of 610, across 29 files
- implement one item at a timein 45 of 610, across 11 files
- group findings by severityin 44 of 610, across 43 files
- verify feedback against the codebasein 42 of 610, across 8 files
- dispatch a code reviewer subagentin 39 of 610, across 23 files
- fix important issues before proceedingin 37 of 610, across 22 files
- test each fix individuallyin 35 of 610, across 7 files
- reply in github comment threadsin 33 of 610, across 5 files
- check for security vulnerabilitiesin 31 of 610, across 27 files
- factualize corrections without over-explainingin 30 of 610, across 2 files
Said here and by no other author read
- read acceptance criteria and the diff
- assess clarity and maintainability
- separate confirmed defects from suggestions
- block on confirmed correctness or security defects
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.