Menu
Spawn 3 axis-specific reviewers (A/B/C)
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Install Pasture binaries (pastured, pasture, pasture-release) from GitHub Releases, go install, or Nix
Pasture protocol reference documentation — 12-phase workflow, agent roles, constraints, and coding standards. Read when you need to understand the full workflow or look up conventions.
Create handoff document and transfer to supervisor
Create PROPOSAL-N task with full technical plan
Ratify proposal, mark old proposals pasture:superseded
Specification writer and implementation designer
| name | architect-request-review |
| description | Spawn 3 axis-specific reviewers (A/B/C) |
Command: pasture:architect:request-review — Spawn 3 axis-specific reviewers (A/B/C)
-> Full workflow in PROCESS.md <- Phase 4
[arch-review-spawn-3-axes]
[arch-review-provide-context]
Plan draft complete, ready for review.
Reviews are named PROPOSAL-N-REVIEW-{axis}-{round} where:
| Axis | Focus | Key Questions |
|---|---|---|
| A | Correctness (spirit and technicality) | Does it faithfully serve the user? Are technical decisions consistent with rationale? |
| B | Test quality | Integration over unit? SUT not mocked? Shared fixtures? Assert outcomes? |
| C | Elegance and complexity matching | Right API? Not over/under-engineered? Complexity proportional to problem? |
Task(description: "Reviewer A: correctness", prompt: "Review PROPOSAL-1 task <task-id>. URD: <urd-id> (read for requirements context). You are Reviewer A (Correctness). Focus: Does it faithfully serve the user? Are technical decisions consistent with rationale? Create review task titled PROPOSAL-1-REVIEW-A-1...", subagent_type: "general-purpose")
Task(description: "Reviewer B: test quality", prompt: "Review PROPOSAL-1 task <task-id>. URD: <urd-id> (read for requirements context). You are Reviewer B (Test quality). Focus: Integration over unit? SUT not mocked? Shared fixtures? Assert outcomes? Create review task titled PROPOSAL-1-REVIEW-B-1...", subagent_type: "general-purpose")
Task(description: "Reviewer C: elegance", prompt: "Review PROPOSAL-1 task <task-id>. URD: <urd-id> (read for requirements context). You are Reviewer C (Elegance). Focus: Right API? Not over/under-engineered? Complexity proportional to problem? Create review task titled PROPOSAL-1-REVIEW-C-1...", subagent_type: "general-purpose")
All 3 reviewers must vote ACCEPT. Max revision rounds until consensus.
bd show <proposal-id>
bd comments <proposal-id>
# Add comment to notify that review is ready
bd comments add <proposal-id> "Review requested — 3 reviewers spawned"
# Check for review votes
bd comments <proposal-id>
For FOLLOWUP_PROPOSAL-N reviews, use the same procedure:
FOLLOWUP_PROPOSAL-N-REVIEW-{axis}-{round}