What is the difference between Path A and Path B verdicts?

Path A signifies full achievement of all criteria, while Path B indicates completion with acceptable external blockers that were out of the agent's control.

Does this skill run existing project tests?

Yes, it automatically detects common test configurations (like npm, pytest, or go test) and runs them independently to verify success claims.

How does it handle vague success criteria?

The validator is programmed to reject unmeasurable criteria, such as 'make progress' or 'improve code,' requiring measurable metrics or specific evidence to pass.

When is the Bazinga Validator skill triggered?

It is invoked only when a Project Manager agent sends a 'BAZINGA' signal, indicating that the task or project is claimed as complete and ready for final review.

Can the validator detect if an agent skipped part of the work?

Yes, it performs a mandatory scope validation by comparing the final output against the original user request and the session database to flag unauthorized scope reduction.

Bazinga Validator

Name: Bazinga Validator
Author: mehdic

bymehdic

•

安全与测试

Validates multi-agent project completion claims through independent verification of tests, criteria, and scope.

Bazinga Validator acts as the final quality gate for autonomous multi-agent development teams using Claude Code. When a Project Manager agent claims completion, this skill steps in to provide a brutally skeptical, independent audit. It verifies test results, parses coverage reports, checks numeric and boolean success criteria, and ensures the original project scope hasn't been reduced without authorization. By requiring concrete evidence and executing its own verification logic, it prevents premature project sign-offs and ensures high-quality deliverables in automated workflows.

主要功能

01Automated verification of success criteria from the Bazinga database

02Structured verdict reporting (ACCEPT/REJECT) for agent orchestration

03Scope mismatch detection and mandatory scope validation

04Evidence-based parsing for coverage, performance, and build status

051 GitHub stars

06Independent test execution with configurable timeouts

使用场景

01Verifying that code coverage meets specific thresholds independently of the developer agent

02Auditing complex multi-agent workflows to ensure all user-requested features were implemented

03Ensuring 100% test pass rates before an AI agent marks a task as done

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mehdic/bazinga bazinga-validator

For use in Claude.ai and ChatGPT

Download Skill