Menu
Pre-PR verification loop. Runs build, type check, lint, security audit, and debug statement scan. Use before creating PRs or merging.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Run the build and intelligently fix errors with guardrails. Stops if fixes introduce more errors or the same error persists after 3 attempts.
Security and quality review of uncommitted changes. Checks for vulnerabilities, code smells, and best practice violations. Use before committing.
Context window conservation rules. Invoke when approaching context limits or before large tasks.
Create session learning logs that persist institutional memory across Claude Code sessions.
Deep reflection on the skill learning system itself. Analyzes what's working, what's stale, and proposes structural improvements. The meta-skill.
Deep security audit for web applications. Checks secrets, input validation, XSS, API security, and dependency vulnerabilities.
| name | verify |
| updated | "2026-02-20T00:00:00.000Z" |
| description | Pre-PR verification loop. Runs build, type check, lint, security audit, and debug statement scan. Use before creating PRs or merging. |
| argument-hint | ["quick|full|pre-pr"] |
| (default | full) |
| allowed-tools | Bash, Grep, Glob, Read, TodoWrite |
Comprehensive verification before PRs. Runs all quality gates and produces a structured pass/fail report.
Check the argument. Default to full if none provided.
Detect the build system:
package.json → npm/yarn/pnpm (check for build, typecheck, lint scripts)Cargo.toml → cargo (build, clippy)go.mod → go (build, vet, lint)pyproject.toml → python (build, mypy, ruff/flake8)Makefile → make targetsRun checks sequentially. Stop early on CRITICAL failures (build, types) since later checks depend on them.
<build-command> 2>&1
If build fails, STOP. Report the errors and suggest running /build-fix.
The build usually includes type checking. For explicit type check without full build:
npx tsc --noEmitmypy . or pyrightgo vet ./...cargo checkJS/TS: npm run lint or npx eslint .
Python: ruff check . or flake8
Go: golangci-lint run
Rust: cargo clippy
PASS: No lint errors (warnings are OK)
FAIL: Any lint errors
Search for debug statements in source files (excluding node_modules, build output, tests):
| Language | Pattern |
|---|---|
| JS/TS | console.log( |
| Python | print( (in non-CLI code), breakpoint(), pdb.set_trace() |
| Go | fmt.Println( (debug-only) |
| Rust | dbg!( |
Search for potential secrets and sensitive data:
Use Grep tool to search across source files:
api[_-]?key\s*[:=]\s*['"][^'"]+['"] (case insensitive)secret\s*[:=]\s*['"][^'"]+['"] (case insensitive)password\s*[:=]\s*['"][^'"]+['"] (case insensitive)-----BEGIN.*PRIVATE KEY-----.env files committed: git ls-files | grep '\.env'Also check that .env is in .gitignore.
git status
git diff --stat
Report uncommitted changes, untracked files.
VERIFICATION: [PASS/FAIL]
═══════════════════════════════
Build: [PASS/FAIL]
Types: [PASS/FAIL] (X errors)
Lint: [PASS/FAIL] (X issues)
Debug stmts: [CLEAN/X found]
Security: [CLEAN/X issues] (pre-pr only)
Git: [clean/X uncommitted changes]
═══════════════════════════════
Ready for PR: [YES/NO]
If any check fails, list the specific failures below the report with file:line references.
If verification fails:
/build-fix