irfanfaraaz/claude-kit/feature-verify

Feature Verify Skill

Purpose

After you code a feature with /coding-agent, you can verify it works using this skill. It provides three testing depths so you control the testing overhead: quick health check, targeted feature test, or full regression suite.

Usage

After implementing a feature (or during /coding-agent if you chose testing option A):

/feature-verify "Feature name" mode:quick

Replace "Feature name" with the feature from features.json (e.g., "Login System", "User Settings").

Testing Modes

Quick Mode (30 seconds)

/feature-verify mode:quick

• Runs init.sh health check only
• Verifies app starts without errors
• Quick smoke test — no feature-specific tests
• Use when: You want fast feedback that the app isn't broken

Example output:

✅ Health Check: PASSED
- App starts: OK
- Dependencies: OK
- Ready to implement more features

Feature Mode (2-5 minutes)

/feature-verify "Login" mode:feature

• Tests the specific feature you just coded
• Runs unit tests + E2E browser tests (Puppeteer MCP)
• Walks through feature steps: load page → interact → verify results
• Use when: You want confidence the feature actually works

Example output:

✅ Login Feature: PASSED
- E2E: User can enter email, password, click submit, redirected to dashboard
- Coverage: 8/8 test cases passed
- Updated features.json: "passes": true

Full Mode (10+ minutes)

/feature-verify mode:full

• Runs entire features.json test suite
• Tests all completed features (regression testing)
• Use when: Before releasing or to catch unexpected breakage

Example output:

✅ Full Test Suite: PASSED (12 of 12)
- Login: ✅ PASSED
- Signup: ✅ PASSED
- Dashboard: ✅ PASSED
- Email Verification: ✅ PASSED
...
- Overall: 12 passed, 0 failed

What It Does

1. Finds the feature in features.json by name
2. Runs appropriate tests:

• quick: init.sh (Puppeteer smoke test)
• feature: Unit tests + feature-specific E2E tests
• full: Entire test suite (all features)

1. Updates features.json:

• Pass: "passes": true
• Fail: "passes": false + logs error details

1. Reports results: ✅ PASS or ❌ FAIL with details

Testing Tools Used

Integration with /coding-agent

When /coding-agent finishes implementing a feature, it asks:

Testing Options:
(A) Run tests → marks passes: true/false
(B) Mark untested → skips tests
(C) Ask later → defer decision
 
Choose (1/2/3):

If you choose (A), the agent runs tests. If you choose (B) or (C), you can use /feature-verify anytime to test later.

Error Handling

If tests fail:

• Feature-verify reports the error
• Updates features.json: "passes": false
• Suggests next steps: fix code, run tests again

If feature name not found:

• Lists available features in features.json
• Ask you to pick from the list