afc-validate

name	afc:validate
description	Validate spec/plan/task artifact consistency
argument-hint	[validation scope: spec-plan, tasks-only]
user-invocable	false
context	fork
allowed-tools	["Read","Grep","Glob"]
model	sonnet

/afc:validate — Artifact Consistency Validation

Validates consistency across spec.md, plan.md, and tasks.md. Read-only — does not modify files.

Project Config (auto-loaded)

!cat .claude/afc.config.md 2>/dev/null || echo "[CONFIG NOT FOUND — run /afc:init first]"

Arguments

• $ARGUMENTS — (optional) limit validation scope (spec-plan, tasks-only)

Execution Steps

1. Load Artifacts

From .claude/afc/specs/{feature}/:

• spec.md (required), plan.md (required)
• tasks.md, research.md, context.md (load if present)

Warn about missing files but proceed with what is available. If config is missing: use CLAUDE.md for architecture info; assume "Layered Architecture" as fallback.

No artifacts early-exit: If both spec.md and plan.md are missing (no feature directory exists or directory is empty), output:

Validation skipped — no pipeline artifacts found
├─ spec.md: not found
├─ plan.md: not found
└─ Run /afc:spec followed by /afc:plan to generate artifacts for validation

Exit without running validation categories. Do not produce a degenerate report with empty coverage data.

2. Run Validation

Run all 6 categories defined in validation-categories.md.

If $ARGUMENTS specifies a scope (e.g., spec-plan), run only the relevant categories.

3. Output Results

## Consistency Analysis Results: {feature name}

### Findings
| ID | Category | Severity | Location | Summary | Recommended Action |
|----|----------|----------|----------|---------|-------------------|
| A-001 | COVERAGE | HIGH | spec FR-003 | No mapping in tasks | Add task |

### Coverage Summary
| Mapping | Coverage |
|---------|----------|
| spec → plan | {N}% |
| plan → tasks | {N}% |
| spec → tasks | {N}% |

### Metrics
- Total requirements: {N} / Total tasks: {N}
- Issues: CRITICAL {N} / HIGH {N} / MEDIUM {N} / LOW {N}

### Next Steps
{Concrete action proposals for CRITICAL/HIGH issues}

4. After Validation Fails

If CRITICAL or HIGH issues are found:

1. Return the findings table to the calling skill (spec/plan/tasks/implement)
2. The calling skill is responsible for fixing the reported issues before proceeding
3. To re-validate after fixes: re-invoke /afc:validate with the same scope

If only MEDIUM/LOW issues: proceed is safe; issues are advisory.

5. Final Output

Analysis complete
├─ Found: CRITICAL {N} / HIGH {N} / MEDIUM {N} / LOW {N}
├─ Coverage: spec→plan {N}%, plan→tasks {N}%, spec→tasks {N}%
└─ Recommended: {next action}

Notes

• Read-only: Report only, no file writes.
• Avoid false positives: Do not over-flag ambiguity — consider context.
• Optional in pipeline: Not required. Can proceed plan → implement directly.