self-critique-verification

Automate design, implementation, and measurement of LangGraph-based self-critique loops.

7|Updated Apr 19, 2026
One-click install
npx skills add https://github.com/namojo/gemini-harness --skill self-critique-verification
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-critique-verification
Source: https://github.com/namojo/gemini-harness/tree/main/.claude/skills/self-critique-verification
Command: npx skills add https://github.com/namojo/gemini-harness --skill self-critique-verification

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Structured design and quantitative validation of self-critique loops within LangGraph-based workflows to reduce defects and improve reliability.

Core Features & Use Cases

  • LangGraph-based loop templates (generate → test → route) with retry and escalation.
  • Structured test runner and fix agent interactions to automatically improve outputs.
  • A/B style evaluation with acceptance criteria to quantify defect reduction and performance.
  • Near-miss trigger tests to validate escalation paths and safe fallbacks.

Quick Start

Configure and run a LangGraph self-critique loop validation in your project.

Frequently Asked Questions about self-critique-verification

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

FAQPage Schema
How do I implement self-critique loops in LangGraph for automated verification?

Self-critique loops in LangGraph automate verification by routing outputs through a structured generate, test, and fix cycle. This workflow enforces retry counters and escalation rules to systematically reduce defects and improve reliability.

What is near-miss testing and how does it validate escalation paths?

Near-miss testing validates escalation paths by applying trigger tests to evaluate safe fallbacks. It quantifies performance during AB-style experiments to ensure robust defect reduction and auditable improvements within automated workflows.

Can I use LangGraph to run AB-style experiments for defect reduction?

Yes, LangGraph supports AB-style evaluation with acceptance criteria to quantify defect reduction. It applies structured test runners and fix agents to automatically improve outputs and measure performance gains.

How do diff-based verdicts work in automated test runners?

Diff-based verdicts in automated test runners evaluate output changes by comparing differences against acceptance criteria. This mechanism ensures structured, auditable improvements by routing results through fix agents and escalation rules.

Does LangGraph support retry counters and escalation rules for fix agents?

LangGraph supports retry counters and escalation rules to manage fix agents effectively. This structured routing ensures outputs automatically improve through enforced loops, validating near-miss triggers and safe fallbacks.