aif-verify（aif 验证）

Verify — Post-Implementation Quality Check

Verify that the completed implementation matches the plan, nothing was missed, and the code is production-ready.

This skill is optional — invoked after /aif-implement finishes all tasks, or manually at any time.

Step 0: Load Context

0.0 Load config.yaml

FIRST: Read .ai-factory/config.yaml if it exists to resolve:

• Paths: paths.description, paths.architecture, paths.rules_file, paths.roadmap, paths.plan, paths.plans, paths.fix_plan, paths.specs, and paths.rules
• verify_mode: default verification strictness (strict | normal | lenient)
• Git: git.enabled, git.base_branch, git.create_branches
• Rules hierarchy: the resolved RULES.md path + rules.base + named rules.<area> entries

verify_mode priority:

1. --strict CLI flag → always use strict
2. config.yaml workflow.verify_mode → use configured value
3. Default → normal

If config.yaml doesn't exist, use defaults:

• Paths: .ai-factory/ for all artifacts
• verify_mode: normal
• Rules: RULES.md only

0.1 Load Ownership and Gate Contract

• Read references/CONTEXT-GATES-AND-OWNERSHIP.md first.
• Treat it as the canonical source for:

- command-to-artifact ownership, - read-only behavior for aif-commit/aif-review/aif-verify, - normal vs strict context-gate thresholds.

• If this contract conflicts with older examples in this file, follow the contract.

0.2 Find Plan File

Same logic as /aif-implement:

1. Check current git branch:
   git branch --show-current
   → Look for <configured plans dir>/<branch-name>.md
2. If the branch-based plan is missing or git mode is off:
   → Check whether the configured plans dir contains exactly one `*.md` full-mode plan
   → If exactly one exists, use it
   → If multiple exist, ask the user to choose or use `@<path>` via `/aif-implement`
3. No full-mode plan → Check the resolved fast plan path
4. No full-mode plan and no resolved fast plan → fall back to standalone verification choices

If no plan file found:

AskUserQuestion: No plan file found. What should I verify?

Options:
1. Verify last commit — Check the most recent commit for completeness
2. Verify branch diff — Compare current branch against the configured base branch
3. Cancel

0.2 Read Plan & Tasks

• Read the plan file to understand what was supposed to be implemented
• TaskList → get all tasks and their statuses
• Read .ai-factory/DESCRIPTION.md (use path from config) for project context (tech stack, conventions)
• Read .ai-factory/ARCHITECTURE.md (use path from config) for dependency and boundary rules (if present)
• Read rules hierarchy (use paths from config):

1. RULES.md — axioms (universal project rules) 2. rules/base.md — project-specific base conventions 3. rules. — area-specific rule entries resolved from config (for example rules.api, rules.frontend)

• Read .ai-factory/ROADMAP.md (use path from config) for milestone alignment checks (if present)

Read .ai-factory/skill-context/aif-verify/SKILL.md — MANDATORY if the file exists.

This file contains project-specific rules accumulated by /aif-evolve from patches, codebase conventions, and tech-stack analysis. These rules are tailored to the current project.

How to apply skill-context rules:

• Treat them as project-level overrides for this skill's general instructions
• When a skill-context rule conflicts with a general rule written in this SKILL.md, the skill-context rule wins (more specific context takes priority — same principle as nested CLAUDE.md files)
• When there is no conflict, apply both: general rules from SKILL.md + project rules from skill-context
• Do NOT ignore skill-context rules even if they seem to contradict this skill's defaults — they exist because the project's experience proved the default insufficient
• CRITICAL: skill-context rules apply to ALL outputs of this skill — including the Verification Report template. If a skill-context rule says "verification MUST check X" or "report MUST include section Y" — you MUST augment the report accordingly. Generating a verification that ignores skill-context rules is a bug.

Enforcement: After generating any output artifact, verify it against all skill-context rules. If any rule is violated — fix the output before presenting it to the user.

0.3 Gather Changed Files

# All files changed during this feature/plan
git diff --name-only <configured-base-branch>...HEAD
# Or if on the base branch / in no-git mode, check recent commits
git diff --name-only HEAD~$(number_of_tasks)..HEAD

If git.enabled = false, skip branch diffing entirely and gather changed files from:

• the working tree (if uncommitted changes exist), or
• the recent commit window that corresponds to the implemented tasks.

Store as CHANGED_FILES.

Step 1: Task Completion Audit

Go through every task in the plan and verify it was actually implemented.

For each task:

1.1 Read Task Description

TaskGet(taskId) → Get full description, requirements, acceptance criteria

1.2 Verify Implementation Exists

For each requirement in the task description:

• Use Glob and Grep to find the code that implements it
• Read the relevant files to confirm the implementation is complete
• Check that the implementation matches what was described, not just that "something was written"

1.3 Build Checklist

For each task, produce a verification result:

✅ Task #1: Create user model — COMPLETE
   - User model created at src/models/user.ts
   - All fields present (id, email, name, createdAt, updatedAt)
   - Validation decorators added

⚠️ Task #3: Add password reset endpoint — PARTIAL
   - Endpoint created at src/api/auth/reset.ts
   - MISSING: Email sending logic (task mentioned SendGrid integration)
   - MISSING: Token expiration check

❌ Task #5: Add rate limiting — NOT FOUND
   - No rate limiting middleware detected
   - No rate-limit related packages in dependencies

Statuses:

• ✅ COMPLETE — all requirements verified in code
• ⚠️ PARTIAL — some requirements implemented, some missing
• ❌ NOT FOUND — implementation not detected
• ⏭️ SKIPPED — task was intentionally skipped by user during implement

Step 2: Code Quality Verification

2.1 Build & Compile Check

Detect the build system and verify the project compiles:

If build fails → report errors with file:line references.

2.2 Test Check

If the project has tests and they were part of the plan:

If tests fail → report which tests failed and whether they relate to the implemented tasks.

If no tests exist or testing was explicitly skipped in the plan → note it but don't fail.

2.3 Lint Check

If linters are configured:

Only lint the changed files to keep output focused.

2.4 Import & Dependency Check

• Verify no unused imports were left behind
• Check that new dependencies mentioned in tasks were actually added (package.json, go.mod, requirements.txt, composer.json)
• Check for missing dependencies (imports that reference packages not in dependency files)

Step 3: Consistency Checks

3.1 Plan vs Code Drift

Check for discrepancies between what the plan says and what was built:

• Naming: Do variable/function/endpoint names match what the plan specified?
• File locations: Are files where the plan said they should be?
• API contracts: Do endpoint paths, request/response shapes match the plan?

3.2 Leftover Artifacts

Search for things that should have been cleaned up:

Grep in CHANGED_FILES: [T][O][D][O]|[F][I][X][M][E]|HACK|[X][X][X]|TEMP|PLACEHOLDER|console\.log\(.*debug|print\(.*debug

Report any found — they might be intentional, but flag them.

3.3 Configuration & Environment

Check if the implementation introduced any new config requirements:

• New environment variables referenced but not documented
• New config files mentioned in code but not created
• Database migrations created but not documented in README/docs

Grep in CHANGED_FILES: process\.env\.|os\.Getenv\(|os\.environ|env\(|getenv\(|config\(

Cross-reference with .env.example, .env.local, README, or docs to ensure they're documented.

3.4 DESCRIPTION.md Sync

Check if .ai-factory/DESCRIPTION.md reflects the current state:

• New dependencies/libraries added during implementation → should be listed
• Architecture changes → should be reflected
• New integrations → should be documented

3.5 Context Gates (Architecture / Roadmap / Rules)

Apply the canonical contract from references/CONTEXT-GATES-AND-OWNERSHIP.md.

Evaluate and report each gate explicitly:

• Architecture gate

- Pass: implementation follows documented boundaries and dependency rules - Warn: architecture mapping is ambiguous or stale - Fail: clear violation of explicit architecture constraints

• Rules gate

- Pass: implementation follows explicit project rules - Warn: relevance/verification is ambiguous - Fail: clear violation of explicit rule text

• Roadmap gate

- Pass: work aligns with existing milestone direction (prefer ## Roadmap Linkage from the plan when present) - Warn: .ai-factory/ROADMAP.md missing, ambiguous mapping, or no milestone linkage for feat/fix/perf scope - Fail (strict mode): clear roadmap contradiction after all available roadmap context is considered

Normal mode behavior:

• Architecture/rules clear violations fail verification.
• Roadmap mismatch and missing milestone linkage are warnings unless contradiction is explicit and severe.

Strict mode behavior:

• Architecture and rules clear violations fail verification.
• Clear roadmap mismatch fails verification.
• Missing milestone linkage for feat/fix/perf remains a warning (even when .ai-factory/ROADMAP.md exists).

Logging/reporting format:

• Non-blocking findings: WARN [gate-name]...
• Blocking findings: ERROR [gate-name]...

3.6 Context Drift (Optional Remediation)

/aif-verify is read-only for context artifacts. Do not edit or regenerate .ai-factory/* files here.

If you detect that a context artifact is stale, missing, or ambiguous, report it as a drift finding and provide the owner-command remediation:

• DESCRIPTION.md drift → suggest /aif (or note that /aif-implement should have updated it during implementation)
• ARCHITECTURE.md drift → suggest /aif-architecture
• ROADMAP.md drift → suggest /aif-roadmap check (or /aif-roadmap <update request>)
• RULES.md drift → suggest /aif-rules <rule text>

Ask the user a single optional question only if drift was detected and fixing it now would materially improve correctness:

AskUserQuestion: Context drift detected. Capture updates now?

Options:
1. Yes — show the exact commands to run (recommended)
2. No — proceed without updating context

Step 4: Verification Report

4.1 Display Results

## Verification Report

### Task Completion: 7/8 (87%)
| # | Task | Status | Notes |
|---|------|--------|-------|
| 1 | Create user model | ✅ Complete | |
| 2 | Add registration endpoint | ✅ Complete | |
| 3 | Add password reset | ⚠️ Partial | Missing: email sending |
| 4 | Add JWT auth middleware | ✅ Complete | |
| 5 | Add rate limiting | ✅ Complete | |
| 6 | Add input validation | ✅ Complete | |
| 7 | Add error handling | ✅ Complete | |
| 8 | Update API docs | ❌ Not found | No changes in docs/ |

### Code Quality
- Build: ✅ Passes
- Tests: ✅ 42 passed, 0 failed
- Lint: ⚠️ 2 warnings in src/api/auth/reset.ts

### Issues Found
1. **Task #3 incomplete** — Password reset endpoint created but email sending not implemented (SendGrid integration missing)
2. **Task #8 not done** — API documentation not updated despite plan requirement
3. **2 unfinished markers found** — src/services/auth.ts:45, src/middleware/rate-limit.ts:12
4. **New env var undocumented** — `SENDGRID_API_KEY` referenced but not in .env.example

### No Issues
- All imports resolved
- No unused dependencies
- DESCRIPTION.md up to date
- No leftover debug logs

4.2 Determine Overall Status

• All Green — everything verified, no issues
• Minor Issues — small gaps that can be fixed quickly
• Significant Gaps — tasks missing or partially done, needs re-implementation

4.3 Action on Issues

If issues were found:

AskUserQuestion: Verification found issues. What should we do?

Options:
1. Fix now (recommended) — Use /aif-fix to address all issues
2. Fix critical only — Use /aif-fix for incomplete tasks, skip warnings
3. Fix directly here — Address issues in this session without /aif-fix
4. Accept as-is — Mark everything as done, move on

If "Fix now" or "Fix critical only":

• First suggest using /aif-fix and pass a concise issue summary as argument
• Example:

- /aif-fix complete Task #3 password reset email flow, implement Task #8 docs update, remove unfinished markers in src/services/auth.ts and src/middleware/rate-limit.ts, document SENDGRID_API_KEY in.env.example

• If user agrees, proceed via /aif-fix
• If user declines /aif-fix, continue with direct implementation in this session
• For each incomplete/partial task — implement the missing pieces (follow the same implementation rules as /aif-implement)
• For unfinished markers/debug artifacts — clean them up
• For undocumented config — update .env.example and docs
• After fixing, re-run the relevant verification checks to confirm

If "Accept as-is":

• Note the accepted issues in the plan file as a comment
• Continue to Step 5

Step 5: Suggest Follow-Up Skills

After verification is complete, suggest next steps based on result:

• If unresolved issues remain (accepted or deferred), suggest /aif-fix first
• If all green, suggest security/review/commit flow

## Verification Complete

Suggested next steps:

1. 🛠️ /aif-fix [issue summary] — Fix remaining verification issues
2. 🔒 /aif-security-checklist — Run security audit on the new code
3. 👀 /aif-review — Code review of the implementation
4. 💾 /aif-commit — Commit the changes

Which would you like to run? (or skip all)

AskUserQuestion: Run additional checks?

Options:
1. Fix issues — Run /aif-fix with verification findings
2. Security check — Run /aif-security-checklist on changed files
3. Code review — Run /aif-review on the implementation
4. Both — Run security check, then code review
5. Skip — Proceed to commit

If fix issues selected → suggest invoking /aif-fix <issue summary> If security check selected → suggest invoking /aif-security-checklist If code review selected → suggest invoking /aif-review If both → suggest security first, then review If skip → suggest /aif-commit

Context Cleanup

Suggest the user to free up context space if needed: /clear (full reset) or /compact (compress history).

Strict Mode

When invoked with --strict:

/aif-verify --strict

• All tasks must be COMPLETE — no partial or skipped allowed
• Build must pass — fail verification if build fails
• Tests must pass — fail verification if any test fails (tests are required in strict mode)
• Lint must pass — zero warnings, zero errors
• No unfinished task markers in changed files
• No undocumented environment variables
• Architecture gate must pass — fail on clear boundary/dependency violations
• Rules gate must pass — fail on clear rule violations
• Roadmap gate must pass — fail on clear roadmap mismatch
• Missing milestone linkage for feat/fix/perf is a warning even in strict mode
• Do not fail strict verification solely because milestone linkage is missing

Strict mode is recommended before merging to the configured base branch or creating a pull request.

Usage

After implement (suggested automatically)

/aif-verify

Strict mode before merge

/aif-verify --strict

Standalone (no plan, verify branch diff)

/aif-verify
→ No plan found → verify branch diff against the configured base branch