critic-review

Reviews Orchestrator decisions after Define stage and returns structured criticism before Execute.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/oleyna80/agentic-sdlc-framework --skill critic-review-oleyna80
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: critic-review
Source: https://github.com/oleyna80/agentic-sdlc-framework/tree/main/.opencode/skills/critic-review
Command: npx skills add https://github.com/oleyna80/agentic-sdlc-framework --skill critic-review-oleyna80

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It provides an independent, read-only challenge to Orchestrator decisions between the Define and Execute stages, catching scope errors, weak risk assessments, and unjustified routing before implementation begins. ## Core Features & Use Cases - Structured Challenge Points: Examines requirements contradictions, source-of-truth errors, runtime and integration choices, isolation claims, skipped assurance functions, and Hard Stop boundaries. - Four Verdicts: Returns APPROVE, APPROVE_WITH_CHANGES, RECONSIDER, or BLOCKED so the Orchestrator can resolve criticism with a clear decision path. - Trigger-Based Invocation: Activates when Stage 0 preflight completes, side-effect class is standard or above, new subagent topology is proposed, or 2+ skip reasons appear in one Work Block. - Use Case: After planning a multi-file refactor, invoke this review to verify the scope, skip reasons, and verification plan actually prove the acceptance criteria before any source writes occur. ## Quick Start Ask the agent to run a critic review of the completed Stage 0 Define output and return a verdict before starting Stage 1 Execute.

Frequently Asked Questions about critic-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review an AI agent's plan before it writes code?▼

Invoke the critic review after Stage 0 Define completes and before Stage 1 Execute. It challenges scope, routing, and risk assessment read-only, then returns a structured verdict the Orchestrator resolves.

When should a critic review be triggered in an agentic SDLC?▼

Trigger it when Stage 0 preflight is complete, the side-effect class is standard or above, a new subagent topology is proposed, or a Work Block contains two or more skip reasons.

What verdicts does the critic review return?▼

It returns APPROVE, APPROVE_WITH_CHANGES, RECONSIDER, or BLOCKED. The Critic only provides structured criticism; the Orchestrator makes the final decision on how to proceed.

Can the critic review modify files or block execution directly?▼

No. The Critic role is strictly read-only, using git status, diff, log, and show for repository inspection. It does not issue BLOCKED or READY gates itself; the Orchestrator resolves its criticism.

What kinds of planning errors does the critic review catch?▼

It catches missing or contradictory requirements, scope and source-of-truth errors, unjustified runtime or integration choices, weak isolation claims, skipped assurance functions, and verification plans that cannot prove acceptance criteria.