Afc:test

/afc:test — Test Strategy Planning and Test Writing

> Establishes a test strategy for implemented code and writes tests. > Standalone command — not part of the auto pipeline. Use after implement or before review.

Arguments

• $ARGUMENTS — (required) Test target. One of:
• File path or directory (e.g., src/features/auth/)
• Feature name (e.g., login flow)
• coverage — full coverage analysis and gap remediation

Project Config (auto-loaded)

!cat .claude/afc.config.md 2>/dev/null || echo "[CONFIG NOT FOUND] .claude/afc.config.md not found. Create it with /afc:init."

Values used: {config.gate} (CI), {config.architecture} (layers), test framework from Project Context.

Execution Steps

1. Target Analysis

1. Parse $ARGUMENTS:

• File/Directory → read that code, check for existing test files
• Feature name → explore related files, read code
• coverage → scan all existing tests, identify coverage gaps

1. Existing coverage check: Before writing new tests, evaluate what already exists:

• Are there existing test files for the target? What do they cover?
• If existing tests already provide adequate coverage → report: "Existing tests in {path} already cover the core scenarios. No additional tests needed." Ask user: "(1) Add edge case tests only (2) Rewrite tests (3) Abort"
• If partially covered → identify specific gaps and target only those

1. Determine characteristics of target code:

• Public interface (function signatures, component props)
• Dependencies (external APIs, DB, state management)
• Branch points (conditionals, error handling)
• Edge cases

2. Test Strategy Planning

### Test Strategy
- Target: {file/feature}
- Framework: {test framework from Project Context}
- Test types:
  - [ ] Unit tests: {list of target functions/methods}
  - [ ] Integration tests: {component interactions}
  - [ ] E2E tests: {user scenarios} (only if applicable)
- Mocking strategy: {mocking approach per dependency}

Confirm strategy with user before proceeding.

3. Write Tests

Principles:

• AAA pattern: Arrange → Act → Assert
• Test names: it('should {expected behavior} when {condition}') format
• Independence: each test can run independently
• Readability: test code serves as documentation

Priority:

1. Happy path (normal behavior)
2. Error cases (error handling)
3. Edge cases (boundary values)
4. Regression guards (prevent bug recurrence)

Test file location: follows project convention

• Co-location: {filename}.test.{ext} (same directory)
• Separate: __tests__/{filename}.test.{ext} or tests/ directory

4. Critic Loop

> Always read ${CLAUDE_SKILL_DIR}/../../docs/critic-loop-rules.md first and follow it.

Run the critic loop until convergence. Safety cap: 5 passes.

| Criterion | Validation | |-----------|------------| | COVERAGE | Are all core logic and branch points covered? | | QUALITY | Do tests validate behavior, not implementation details? Are there any brittle tests? |

Follow verdict handling and output format per docs/critic-loop-rules.md.

5. Run and Verify Tests

{config.gate}

On failure:

1. Determine whether the issue is in test code or implementation code
2. Test code issue → fix tests
3. Implementation code issue → report to user (test found a bug)

Maximum 3 retries.

6. Final Output

Tests complete
├─ Target: {file/feature}
├─ Written: {N} tests ({unit X, integration Y, E2E Z})
├─ Coverage: {summary of key branch point coverage}
├─ Critic: converged ({N} passes, {M} fixes, {E} escalations)
├─ Verified: all tests passing
└─ Found: {bug details if found, otherwise "no issues"}

Notes

• No implementation coupling: test external behavior, not internal implementation. Do not directly test private methods.
• Minimize mocking: mock only necessary external dependencies. Excessive mocking reduces test value.
• Avoid snapshot overuse: UI snapshots for core structure only. Avoid snapshots that break on style changes.
• Respect existing tests: follow existing test patterns and conventions. Confirm with user before introducing new patterns.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: jhlee0409
• Source: jhlee0409/all-for-claudecode
• License: MIT
• Homepage: https://www.npmjs.com/package/all-for-claudecode

Install and usage instructions live in the source repository linked above.