Specwright Gate: Spec Compliance

Verifies that every acceptance criterion in the spec has corresponding implementation and test evidence.

Default verdict is FAIL. Evidence must be cited before any verdict. Absence of evidence is evidence of non-compliance.

Step 1: Read Configuration and State

Read .specwright/config.json for integration.omc (agent delegation mode). Read .specwright/state/workflow.json to get current epic and specDir. If no epic active, STOP: "No active epic. Run /specwright:specify first."

Step 2: Load Spec

Read .specwright/epics/{specDir}/spec.md and extract all acceptance criteria. Parse lines matching - [ ] or - [x] under "Acceptance Criteria" headings. Build a numbered list of all criteria. If spec.md has zero acceptance criteria: write ERROR status, STOP.

Step 3: Map Criteria to Evidence

Delegate to code-reviewer agent for thorough mapping.

Agent Delegation:

Read .specwright/config.json integration.omc to determine delegation mode.

If OMC is available (integration.omc is true): Use the Task tool with subagent_type:

code

subagent_type: "oh-my-claudecode:code-reviewer"
description: "Map spec to evidence"
prompt: |
  Map acceptance criteria to implementation evidence.

  Acceptance Criteria:
  {numbered criteria list}

  Project root: {cwd}
  Changed files: {git diff --name-only}

  For EACH criterion, provide:
  1. Status: PASS / FAIL / WARN
  2. Test name and file:line that verifies it
  3. Implementation file:line reference
  4. Reason if FAIL or WARN

  Rules:
  - Criterion with no corresponding test = FAIL
  - Criterion with test that doesn't actually assert the behavior = WARN
  - Criterion with clear test coverage = PASS

If OMC is NOT available (standalone mode): Use the Task tool with model:

code

prompt: |
  Map acceptance criteria to implementation evidence.

  Acceptance Criteria:
  {numbered criteria list}

  Project root: {cwd}
  Changed files: {git diff --name-only}

  For EACH criterion, provide:
  1. Status: PASS / FAIL / WARN
  2. Test name and file:line that verifies it
  3. Implementation file:line reference
  4. Reason if FAIL or WARN

  Rules:
  - Criterion with no corresponding test = FAIL
  - Criterion with test that doesn't actually assert the behavior = WARN
  - Criterion with clear test coverage = PASS
model: "opus"
description: "Map spec to evidence"

Step 4: Parse Evidence Response

Extract mapping from code-reviewer response. Count PASS / FAIL / WARN statuses.

Step 5: Baseline Check

If .specwright/baselines/gate-spec.json exists, load entries ({finding, file, reason, expires} with ISO dates; null = no expiry). For matching findings: downgrade BLOCK->WARN, WARN->INFO. Ignore expired entries. Partial match (same category, different line): AskUserQuestion. Log all downgrades in evidence.

Step 6: Update Gate Status

Self-critique checkpoint: Before finalizing — did I accept anything without citing proof? Did I give benefit of the doubt? Would a skeptical auditor agree? Gaps are not future work. TODOs are not addressed. Partial implementations do not match intent. If ambiguous, FAIL.

Determine final status:

•Incomplete analysis: ERROR (invoke AskUserQuestion)
•Any criterion FAIL: FAIL
•Any criterion WARN (test doesn't assert behavior): WARN
•All PASS: PASS

Update .specwright/state/workflow.json gates.spec:

json

{
  "status": "PASS|WARN|FAIL|ERROR",
  "lastRun": "<ISO-timestamp>",
  "evidence": "{specDir}/evidence/spec-compliance.md",
  "verified": {pass_count},
  "unverified": {fail_count}
}

Step 7: Write Evidence Report

Create {specDir}/evidence/spec-compliance.md:

markdown

# Spec Compliance Report
Generated: {timestamp}

## Summary
- Total Criteria: {total}
- Verified (PASS): {pass_count}
- Unverified (FAIL): {fail_count}
- Warnings: {warn_count}

## Criteria Mapping

### Criterion 1: {description}
- Status: PASS
- Test: {test name} ({file}:{line})
- Implementation: {file}:{line}

### Criterion 2: {description}
- Status: FAIL
- Reason: No test found verifying this behavior

Step 8: Output Result

code

SPEC GATE: {PASS|WARN|FAIL}
Criteria: {total} total, {verified} verified, {unverified} unverified
Evidence: {specDir}/evidence/spec-compliance.md

{if FAIL: List unverified criteria}