How is the verification process triggered?

The skill triggers automatically via a post-skill hook whenever parallel subagents finish their tasks, specifically monitoring the transition of running agents to zero.

What specifically does the substance check look for?

It reads generated files to ensure they contain actual content (at least 10 lines) rather than just being empty files or simple boilerplate placeholders.

What happens if a verification check fails?

The skill generates a failure report in the .claude-state directory and provides specific recommendations, such as re-calling the specific subagent or running a build-error-resolver.

Does this skill interrupt my current task?

No, the verifier is designed to run in the background using a lightweight model (Haiku) so that your main orchestration and coding flow remain unblocked.

AI Agent Output Verifier

Name: AI Agent Output Verifier
Author: Wondermove-Inc

byWondermove-Inc

•

보안 및 테스팅

Automatically audits and validates the outputs of parallel AI subagents to ensure code integrity and plan alignment.

The Agent Output Verifier acts as a dedicated quality assurance layer for multi-agent development workflows within Claude Code. It automatically triggers upon the completion of parallel subagent tasks to perform a rigorous five-step verification protocol: checking for file existence, verifying content depth (detecting empty stubs), validating dependency wiring, running TypeScript type checks, and ensuring alignment with the original project plan. By operating in the background, it provides immediate feedback on the success of AI-generated tasks without interrupting the user's primary workflow.

주요 기능

01Automatic background triggering upon parallel subagent completion

021 GitHub stars

03Integrated TypeScript validation using tsc --noEmit

04Five-step verification: Existence, Substance, Wiring, Type Safety, and Plan Alignment

05Deep content inspection to detect empty files or placeholder stubs

06Comprehensive markdown reporting with automated remediation recommendations

사용 사례

01Detecting broken imports and type errors immediately after AI code generation

02Ensuring architectural documentation stays in sync with implementation artifacts

03Validating that complex parallel coding tasks produced complete and functional files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add wondermove-inc/calab-claude-plugin verify-agents

For use in Claude.ai and ChatGPT

Download Skill