Code review
Reviews the changes, identifies critical and high-priority issues, generates review summaries, and collects metrics data for local use. Use this when users need to code review or analyzing code changes for quality issues.From its SKILL.md
npx -y skills add kinneyyan/prompts --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
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 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
6.4 KB, as published. Nobody here has run it
Code Review and Metric Collection
Review the changes and save the metrics data to a local temporary file.
Script Directory
Agent Execution Instructions:
- Determine this SKILL.md file's directory path as
SKILL_DIR - Script path =
${SKILL_DIR}/scripts/<script-name>.sh
| Script | Purpose |
|---|---|
collect-metrics.sh | Collect git statistics and code review metrics to a local temporary file |
show-diff.sh | Show git diff with exclusions |
security-scan.sh | Frontend only — Automated security scan: detects secrets, XSS, eval, hardcoded tokens |
Workflow
Copy this checklist and check off items as you complete them:
- [ ] Step 1: Git Repo verification & Changes confirmation
- [ ] Step 2: Code Review
- [ ] 2.1 Load guidelines
- [ ] 2.2 Security scan (optional — frontend projects only)
- [ ] 2.3 Analyze changes
- [ ] 2.4 Output Result
- [ ] Step 3: User feedback & Validation
- [ ] Step 4: Collect metrics ⚠️ REQUIRED
Step 1: Git Repo verification & Changes confirmation
Ensure if is in a git repository directory and existing changes to review (staged, unstaged, and untracked files):
if ! git rev-parse --git-dir > /dev/null 2>&1; then
echo "Error: Not a git repository. This command requires git version control."
exit 1
fi
echo "Success: Git repository verified."
if [ -z "$(git status --porcelain)" ]; then
echo "No changes detected. Working tree is clean."
exit 1
else
echo "Changes detected."
fi
Step 2: Code Review
2.1 Load guidelines
- Load
references/guidelines.mdas the base set of rules. - Check if
AGENTS.mdexists in the project root. If so, load its content. - If
AGENTS.mdis found, extract rules relevant to code review and merge them into the review standard. - Priority rule: When project conventions conflict with
references/guidelines.md, project conventions override. - If no convention files are found, proceed with base guidelines only — do not fail or warn.
- Record which convention file(s) were loaded and their relevant sections/lines for source attribution in the review output.
2.2 Security scan (optional — frontend projects only)
⚠️ This step is designed for frontend tech stacks (JavaScript / TypeScript / React / Vue / Angular). Skip it for backend, infrastructure, or other non-frontend projects.
If applicable, run automated security detection BEFORE manual review. Results feed into CRITICAL findings in Step 2.3.
bash ${SKILL_DIR}/scripts/security-scan.sh
This detects:
eval()/new Function()— arbitrary code executioninnerHTML/dangerouslySetInnerHTML— XSS vectors- Hardcoded tokens, passwords, API keys
- Sensitive data in
localStorage console.log/console.debugleft in production code- HTTP URLs (non-localhost)
document.write()— blocking + XSS
Any match is automatically classified as CRITICAL or HIGH and included in the final output.
2.3 Analyze changes
- Get staged, unstaged, and untracked changes using Bash:
git status HEAD --short && bash ${SKILL_DIR}/scripts/show-diff.sh` - Analyze code changes to identify issues and store details (FilePath, Line, Suggestion) for subsequent output.
- Set initial counts:
CRITICAL_ISSUES_COUNT,HIGH_PRIORITY_ISSUES_COUNT.
2.4 Output Result
Show categorized issues to the user, exactly follow one of the two templates:
Template A (any findings)
## Code review Skill Output
### 🚨 CRITICAL (Must fix)
#### <brief description of CRITICAL issue>
- FilePath: <path>#<line>
- Suggested fix: <brief description of suggested fix>
- Rule source: "AGENTS.md §<Section/line N>" _(show only when the rule comes from AGENTS.md)_
... (repeat for each CRITICAL issue) ...
---
### ⚠️ HIGH PRIORITY (Should fix)
#### <brief description of HIGH PRIORITY issue>
- FilePath: <path>#<line>
- Suggested fix: <brief description of suggested fix>
- Rule source: "AGENTS.md §<Section/line N>" _(show only when the rule comes from AGENTS.md)_
... (repeat for each HIGH PRIORITY issue) ...
---
### 💡 SUGGESTIONS (Consider)
#### <brief description of SUGGESTIONS issue>
- FilePath: <path>#<line>
- Suggested fix: <brief description of suggested fix>
- Rule source: "AGENTS.md §<Section/line N>" _(show only when the rule comes from AGENTS.md)_
... (repeat for each SUGGESTIONS issue) ...
Template B (no issues)
## Code review Skill Output
🎉 Great! No issues found.
Step 3: User feedback & Validation
- If any issues found, ask the user: "Do you have any objections to the review results?" And two options are provided:
1. Yes, I have objections、2. No, proceed to the next step - If the user choose "Yes":
- Provide a numbered list of all CRITICAL and HIGH PRIORITY issues.
- Ask the user: "Please select the numbers of the issues you disagree with and briefly explain why (e.g., '1-this is a test file, 3-intentional technical debt')"
- Parse the user's response to identify which specific CRITICAL and HIGH PRIORITY issues were rejected.
- Extract the user's reason for each rejected item.
- Calculate
REJECTED_CRITICAL_COUNTandREJECTED_HIGH_COUNT. - Format the rejected details into a single-line string
REJECTION_DETAILS(Format:[Level]Issue brief=>User reason;[Level]Issue brief=>User reason).
- If no issues found or no objections:
REJECTED_CRITICAL_COUNT=0REJECTED_HIGH_COUNT=0REJECTION_DETAILS=""
Step 4: Collect metrics ⚠️ REQUIRED
bash ${SKILL_DIR}/scripts/collect-metrics.sh \
"<Brief summary of changes (single-line only, within 100 words)>" \
"<Detailed review summary (single-line only)>" \
"$CRITICAL_ISSUES_COUNT" \
"$HIGH_PRIORITY_ISSUES_COUNT" \
"$REJECTED_CRITICAL_COUNT" \
"$REJECTED_HIGH_COUNT" \
"$REJECTION_DETAILS"
Notes
- Never modify the code after review