skills.rest
tipmaster
tipmasterCOMMUNITY
Explore all skills by this author →
GitHub

convergence-debugger

Diagnose AI deliberation convergence issues by analyzing transcripts and configuration.

Updated Aug 23, 2026
Search Tags:#machine learning#nlp#convergence#semantic similarity#early stopping#ai debugging#deliberation
One-click install
npx skills add https://github.com/tipmaster/claude-config --skill convergence-debugger-tipmaster
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: convergence-debugger
Source: https://github.com/tipmaster/claude-config/tree/main/mcp-servers/ai-counsel/.claude/skills/convergence-debugger
Command: npx skills add https://github.com/tipmaster/claude-config --skill convergence-debugger-tipmaster

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires sentence-transformers, scikit-learn.

What problem does it solve?

This Skill helps diagnose why AI deliberation systems aren't converging as expected, preventing wasted API calls, improving early stopping mechanisms, and ensuring accurate consensus detection. It provides a systematic workflow for troubleshooting.

Core Features & Use Cases

  • Transcript Analysis: Guides examination of deliberation transcripts for convergence information, voting outcomes, and similarity scores.
  • Configuration Validation: Checks critical settings like semantic_similarity_threshold, min_rounds_before_check, and early_stopping parameters.
  • Backend Diagnostics: Identifies issues with semantic similarity backends (e.g., Jaccard vs. SentenceTransformer) and their impact on scores.
  • Use Case: Diagnose why an AI deliberation is running for its maximum rounds despite participants giving seemingly identical responses, checking configuration, semantic similarity scores, and the chosen backend.

Quick Start

Use the convergence-debugger skill to analyze the latest deliberation transcript and config.yaml to understand why early stopping isn't triggering.