┌──────────────────────────────────────────────────────────────────────┐ │ │ │ ░█▀▀░░█▀█░░█▀█ Code Quality Skills Kit · v3.6 │ │ ░█░░░░█░█░░█░█ ──────────────────────────────────── │ │ ░▀▀▀░░▀▀▀░░▀▀▀ Multi-CLI · 15 dimensions · 99% coverage │ │ Critical-Cap scoring · token-optimised │ │ │ │ ▸ One prompt · Any codebase · Parallel sub-agents │ │ ▸ github.com/servas-ai/code-quality-skills-kit · MIT │ │ │ └──────────────────────────────────────────────────────────────────────┘
Multi-Agent Code Quality Audit Kit · v3.6
One prompt. Any codebase. Parallel sub-agents across Claude, Gemini, OpenCode, Codex.
- 15
- Dimensions
- 4
- CLIs Supported
- 99%
- Coverage Gate
- MIT
- License
CLI Status & Budget Controls
Live placeholder dashboard showing how the kit reports detected CLIs, plan tier, weekly budget, and scheduled resets. All values shown here are demo data — drop in your repo to see the real numbers.
Budget Controls
| Enable | CLI | Status | Account | Plan | Day used | Week used | Resets in | Skills |
|---|
Daily reset
--h --m --s
Next 00:00 UTC
Weekly reset
--d --h --m
Monday 00:00 UTC
Active CLIs
4 / 4
checked rows above
You selected: 3 of 4 CLIs · 75% week budget · daily cap $50 · stop at 90%
Skills will fan out to: claude (d4 d6 d13) · gemini (d12 d8 d9) · codex (d2 d5 d7 d10)
$15 Audit Dimensions
Each CLI plays to its strengths and the dispatcher above fans the 15 dimensions out in parallel. Click any dimension to expand its checks. Filter by audit phase to focus on what matters this iteration.
d1 Correctness & Invariants w15
Catches the bugs that ship.
- Race conditions in async flows
- Swallowed catches (empty
catch {}) as anyescape hatches- Broken project-defined invariants
d3 Tests w12
Validates that tests actually run and cover changes.
- Pass / fail counts from real runs
.skip/.todo/.onlyinventory- Missing tests for changed modules
d6 Architecture w10
Keeps the codebase maintainable as it grows.
- God-files (>300 LOC)
- Layer violations
- Circular dependencies
- Barrel-file hygiene
d8 Dead code & Tech debt w8
Surfaces what should be deleted.
- Unused exports across the graph
- Stale TODOs (>30 days)
- Placeholder files / scaffold leftovers
d2 Type Safety w12
Trust the compiler, then check what it can’t.
- Real
tscerror counts (not estimates) - Missing strict flags
anyinventory@ts-ignoredebt
d10 CI Health w5
What does CI actually say right now?
- Exit codes for typecheck / test / build / lint
- GitHub Actions success rate (last N runs)
d12 Dependencies w6
Audits the supply chain you actually depend on.
pnpm audit/ equivalent- Outdated & duplicate packages
- License risk surface
d5 Performance w10
The slow parts, ranked.
- First Load JS budget
useEffectheat map- N+1 query patterns
- Unscoped imports & missing memoization
d13 Cache keys w8
The most common source of stale UI.
- React-Query key-factory consistency
- Invalidation drift across mutations
staleTimeaudit
d7 Accessibility w10
Keyboard users, screen readers, and contrast.
<div onClick>without keyboard handler- Missing focus rings
- Alt text & form labels
d9 Documentation w6
Docs are a feature; drift is a bug.
- README freshness vs
main - Scope drift between README & code
- JSDoc on public API
- Broken markdown links
d11 i18n w6
Only runs when an extraction policy is active.
- Hardcoded UI strings escaping the catalog
d14 CSS tokens w4
Tokens win over ad-hoc values, every time.
- Tailwind arbitrary values
[#abc] - Raw hex outside the token layer
- Dark-mode parity gaps
d4 Security & Privacy w12
The patterns that get apps owned.
- Hardcoded secrets & API keys
eval,innerHTMLsinks- Env var leaks to client bundle
.gitignorehygiene
d15 Feature flags w4
Old flags are quiet liabilities.
- Stale flags (>90 days)
- Half-rolled-out flags
- Dead branches behind permanently-off flags
Install, configure, and watch it run
Three install paths. Four config questions. One audit you can replay below.
Install 3 ways
curl -sSL https://raw.githubusercontent.com/servas-ai/code-quality-skills-kit/main/install.sh | sh
git clone --depth 1 https://github.com/servas-ai/code-quality-skills-kit /tmp/cqsk cp /tmp/cqsk/code-quality-checker.md ./ cp -r /tmp/cqsk/code-quality-skills ./ echo "audit-reports/" >> .gitignore
# Drop into ~/.claude/commands/ for global slash command mkdir -p ~/.claude/commands curl -sSL https://raw.githubusercontent.com/servas-ai/code-quality-skills-kit/main/code-quality-checker.md \ -o ~/.claude/commands/code-quality-checker.md