How can I improve the accuracy of my custom agents?

Use /routing-debug keywords to review existing triggers and /routing-debug trace to understand how the orchestrator parses your prompts and weighs specific terms.

What factors influence the agent routing confidence score?

Confidence scores are calculated based on weighted keyword matches, specific file patterns, and error patterns defined in the agent configuration files.

Why did the wrong agent respond to my request?

You can use the /routing-debug command to see exactly which keywords or file patterns triggered the selection and compare the confidence scores against other competing agents.

Can I test routing logic without executing a full task?

Yes, the /routing-debug command allows you to simulate routing for any prompt, providing a detailed trace of the decision-making process without running the actual agent tools.

Agent Routing Debugger

Name: Agent Routing Debugger
Author: jrc1883

byjrc1883

0•

Analytics & Monitoring

Diagnoses and optimizes agent selection logic by analyzing keyword matches, confidence scores, and routing decision trees.

The Agent Routing Debugger provides deep visibility into the decision-making process of the Claude Code multi-agent orchestrator. It helps developers understand exactly why specific agents are triggered by identifying keyword matches, file pattern associations, and error pattern triggers. Whether you are dealing with unexpected agent selection, low confidence scores, or tuning a custom agent's activation triggers, this skill offers the diagnostic tools—including routing traces and comparative scoring—needed to refine and optimize your AI-powered development workflows.

Key Features

01Detailed routing analysis with confidence scoring

02Keyword and pattern listing for agent tuning

03Simulated routing traces for prompt testing

04Agent comparison and keyword mapping

05Comprehensive decision-tree visualization

060 GitHub stars

Use Cases

01Optimizing agent activation keywords for custom workflows

02Analyzing low confidence scores to improve prompt engineering or agent configurations

03Troubleshooting why an unexpected agent was selected for a specific prompt

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jrc1883/popkit-claude pop-routing-debug

For use in Claude.ai and ChatGPT

Download Skill