bundled-skills/codebase-audit-pre-push/SKILL.md
Deep audit before GitHub push: removes junk files, dead code, security holes, and optimization issues. Checks every file line-by-line for production readiness.
npx skillsauth add FrancoStino/opencode-skills-antigravity codebase-audit-pre-pushInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
As a senior engineer, you're doing the final review before pushing this code to GitHub. Check everything carefully and fix problems as you find them.
Review the entire codebase file by file. Read the code carefully. Fix issues right away. Don't just note problems—make the necessary changes.
Start by looking for files that shouldn't be on GitHub:
Delete these immediately:
.DS_Store, Thumbs.db, desktop.ini*.log, npm-debug.log*, yarn-error.log**.tmp, *.temp, *.cache, *.swpdist/, build/, .next/, out/, .cache/node_modules/, vendor/, __pycache__/, *.pyc.idea/, .vscode/ (ask user first), *.iml, .project*.bak, *_old.*, *_backup.*, *_copy.*coverage/, .nyc_output/, test-results/TODO.txt, NOTES.txt, scratch.*, test123.*Critical - Check for secrets:
.env files (should never be committed)password, api_key, token, secret, private_key*.pem, *.key, *.cert, credentials.json, serviceAccountKey.jsonIf you find secrets in the code, mark it as a CRITICAL BLOCKER.
Check if the .gitignore file exists and is thorough. If it’s missing or not complete, update it to include all junk file patterns above. Ensure that .env.example exists with keys but no values.
Look through each code file and check:
Dead Code (remove immediately):
return, inside if (false))Code Quality (fix issues as you go):
data, info, temp, thing → rename to be descriptiveif (status === 3) → extract to named constantconsole.log, print(), debuggerany: add proper types or explain why any is used=== instead of == in JavaScriptLogic Issues (critical):
.catch() or try/catchdefault in switch statementsSecrets: Search for hardcoded passwords, API keys, and tokens. They must be in environment variables.
Injection vulnerabilities:
exec() with user-provided inputinnerHTML or dangerouslySetInnerHTML with user dataAuth/Authorization:
Data exposure:
Dependencies:
npm audit or an equivalent toolDatabase:
SELECT *: specify columnsAPI Design:
Code:
Organization:
Separation of concerns:
Reusability:
Backend:
Frontend (if applicable):
README.md must include:
Code comments:
test.only or fdescribe should remain in the codetest.skip without an explanationAfter making all changes, run the app. Ensure nothing is broken. Check that:
After auditing, provide a report:
CODEBASE AUDIT COMPLETE
FILES REMOVED:
- node_modules/ (build artifact)
- .env (contained secrets)
- old_backup.js (unused duplicate)
CODE CHANGES:
[src/api/users.js]
✂ Removed unused import: lodash
✂ Removed dead function: formatOldWay()
🔧 Renamed 'data' → 'userData' for clarity
🛡 Added try/catch around API call (line 47)
[src/db/queries.js]
⚡ Fixed N+1 query: now uses JOIN instead of loop
SECURITY ISSUES:
🚨 CRITICAL: Hardcoded API key in config.js (line 12) → moved to .env
⚠️ HIGH: SQL injection risk in search.js (line 34) → fixed with parameterized query
SCALABILITY:
⚡ Added pagination to /api/users endpoint
⚡ Added index on users.email column
FINAL STATUS:
✅ CLEAN - Ready to push to GitHub
Scores:
Security: 9/10 (one minor header missing)
Code Quality: 10/10
Scalability: 9/10
Overall: 9/10
@security-auditor - Deeper security review@systematic-debugging - Investigate specific issues@git-pushing - Push code after audittools
Authorized security assessment of LLM applications and AI agents: prompt injection, tool abuse, RAG exposure, memory poisoning, system-prompt extraction, and agent-compliance engineering per OWASP LLM/ASI Top 10.
development
Builds two parameterized UI modes—流光溢彩白 (iridescent white) and 五彩斑斓黑 (colorful black)—with OKLCH, WebGL/CSS fallback, vision gating, screenshot QA, and total/per-color intensity reports. Use when a UI request names either mode or needs measured color parameters.
tools
Delegate coding tasks to the Kimi Code CLI (`kimi`) only when the user explicitly requests it, while the orchestrator retains review and landing responsibility.
development
Front-end JavaScript reverse engineering: locate signature chains, analyze encrypted request parameters, sample runtime behavior, and reproduce logic locally in Node for evidence-based output.