production-check

Production Check Skill

Purpose: Verify code is production-ready by running build, lint, and test commands Trigger: After code changes, before deployment or sign-off Input: Working directory, package manager (pnpm) Output: { passed: boolean, build_result, lint_result, test_result, failures[] }

---

Workflow

1. Verify working directory and package manager

• Confirm cwd is monorepo root
• Verify pnpm is available (pnpm --version)
• If package manager not found: return error

2. Run pnpm build

• Execute: pnpm build (or pnpm build --filter=... if scoped)
• Capture stdout/stderr
• Status: ✓ pass or ✗ fail
• If turbo available: runs in parallel (turbo handles dependencies)

3. Run pnpm lint

• Execute: pnpm lint
• Capture stdout/stderr
• Status: ✓ pass or ✗ fail
• Non-blocking if only warnings

4. Run pnpm test

• Execute: pnpm test (or pnpm test --run for CI mode)
• Capture stdout/stderr
• Status: ✓ pass or ✗ fail
• Can fail tests but collect results

5. Aggregate results

• Collect all failures into array
• Return final status (passed = all 3 commands succeeded)
• Provide diagnostic output for fixing

---

Output Format

Success Case:

{
  "passed": true,
  "build_result": {
    "status": "pass",
    "duration": "12s"
  },
  "lint_result": {
    "status": "pass",
    "warnings": 0
  },
  "test_result": {
    "status": "pass",
    "tests": "42 passed"
  },
  "failures": []
}

Failure Case:

{
  "passed": false,
  "build_result": {
    "status": "fail",
    "error": "TypeScript compilation failed",
    "stderr": "src/types.ts:45: error TS2322: Type 'string' not assignable to type 'number'"
  },
  "lint_result": {
    "status": "pass"
  },
  "test_result": {
    "status": "fail",
    "error": "2 tests failed",
    "stderr": "FAIL  src/__tests__/auth.spec.ts"
  },
  "failures": [
    "Build failed: TypeScript compilation",
    "Tests failed: 2 tests in src/__tests__/auth.spec.ts"
  ]
}

---

Error Handling

Scenario	Action
pnpm not found	Return error, ask user to install pnpm
Wrong working dir	Return error, ask user to run from monorepo root
Build fails	Stop, return build error details for fixing
Lint warnings only	Continue (non-blocking), log warnings
Tests fail	Continue (collect results), return test error list
Unknown error	Capture stderr, return for debugging

---

Key Rules

• Root execution: Always run from monorepo root (pwd must show root)
• Turbo optimization: If turbo.json exists, pnpm build uses turbo (automatic)
• Parallel execution: Build runs in parallel (turbo handles), lint and test sequential
• Capture output: All stdout/stderr captured for diagnostics
• No modifications: Read-only execution, no file changes

---

Integration

Called by:

• /build command (validation phase)
• /audit command (quality validation)
• validation-phase skill (before reporting)

Calls: pnpm scripts (no subagents)

Next step: Report results to calling agent for decision

---

Example

Input:

cwd: "/home/user/monorepo"
package_manager: "pnpm"

Execution:

$ cd /home/user/monorepo
$ pnpm build
→ ✓ Build succeeded (12s)

$ pnpm lint
→ ✓ Lint passed

$ pnpm test --run
→ ✓ 42 tests passed

Output:

{
  "passed": true,
  "build_result": { "status": "pass", "duration": "12s" },
  "lint_result": { "status": "pass", "warnings": 0 },
  "test_result": { "status": "pass", "tests": "42 passed" },
  "failures": []
}

---

Configuration

Setting	Value	Rationale
User	root	Monorepo root required
Timeout	5 minutes max	Allow large test suites
Fail-fast	No (collect all)	Show all issues in one pass
Format	JSON output	Machine-readable for agents