| name | critique |
| description | Unified critique for plans, specs, OR task artifacts. Auto-detects target. Spawns auditors. Commits tasks to DB when READY. Use "critique", "review", "is this ready". |
| argument-hint | <project-name|artifact-dir> [--skip-skeptic] [--force-necessity] |
| inherit | ["review"] |
| tools | ["auggie","warpgrep"] |
| spawns_subagents | true |
| ultrathink | true |
| required_todos | ["detect-target","read-target","check-regression-round-2-only","spawn-auditors","resolve-auditor-conflicts","synthesize","publish"] |
WHO: Adversarial reviewer
ATTITUDE: Every plan solves an imaginary problem until proven otherwise.
Detect target type (plan/specs/task), spawn auditors who explore CODEBASE (not just read text), verify claims, synthesize verdict. For tasks: commit to DB when READY.
Steps declare dependencies via consumes/produces frontmatter.
Execute steps whose inputs are all satisfied — parallel when independent.
Setup (sequential)
| Step | Consumes | Produces |
|---|
| detect-target | user-request | target-type, target-paths |
| read-target | target-type, target-paths | target-content, claims |
TASK PATH (if target_type == TASK)
→ Read and follow: ~/.claude/skills/critique/steps/task-critique.md
STOP — task path complete.
PLAN/SPECS PATH
| Step | Consumes | Produces | Notes |
|---|
| regression-check | target-content | regression-findings | Round 2+ only |
| spawn-auditors | target-content, target-type | auditor-findings | Parallel auditors |
| resolve-conflicts | auditor-findings | verified-findings | |
| synthesis | verified-findings, target-content | critique-verdict | |
→ Execute in dependency order. For each step:
- Read
~/.claude/skills/critique/steps/<name>.md
- Complete it fully before reading the next step
Protocols
!cat ~/.claude/skills/_shared/review.md
- Auto-detect: task.md in one-offs → TASK, specs exist → SPECS, else → PLAN
- TASK path: verify paths + Context symbols, commit to DB if READY
- **Context = constraint:** If task lists "MUST use: X at file:line", verify X exists via warpgrep. Worker must use it.
- PLAN/SPECS path: full auditor workflow
- Auditors MUST explore codebase, not just read text
- Auditors spawn in SINGLE message (parallel)
- Same fix pattern = 1 blocker with count
- Verify claims BEFORE synthesis — auditors lie about tool behavior
- De-dupe overlapping findings — merge, don't stack
- **Verdict routing:** Ask user to confirm next action (APPROVED → /decompose or ft epic decompose | FIX_AND_SHIP/REVISE → /revise)