Log and forward LLM API requests to inspect system prompts and tool schemas.
npx skills add https://github.com/JGalego/Bag-of-Tricks --skill snitch-jgalego
Please help me install this Agent Skill. Skill: snitch Source: https://github.com/JGalego/Bag-of-Tricks/tree/main/snitch Command: npx skills add https://github.com/JGalego/Bag-of-Tricks --skill snitch-jgalego
This Skill allows users to inspect the exact requests sent by LLM agents to models, revealing hidden system prompts, tool schemas, and messages, aiding in debugging and auditing.
Run snitch --port 8787 to start the proxy and web UI, then point your SDK to the proxy (e.g., export ANTHROPIC_BASE_URL=http://localhost:8787).
High-intent search queries and answers about installing and using this skill.
To inspect requests sent to LLM APIs, run a local proxy like Snitch to intercept, log, and view the exact system prompts, tool schemas, and messages forwarded to the model. This reveals hidden context for debugging.
Unexpected LLM agent behavior often stems from hidden context. By monitoring API requests with a transparent proxy, you can audit the actual system prompts and tool schemas sent to the model to find discrepancies.
To view hidden system prompts sent to an LLM, route your agent's traffic through a monitoring proxy. This logs the complete request payload, allowing you to inspect and interact with captured data in a live web UI.
Yes, you can debug AI agents by routing their traffic to a local proxy. Point your SDK's base URL to the proxy to monitor and audit the exact LLM API requests, capturing all messages and tool schemas.
Yes, running this LLM API monitoring proxy requires Python 3.9 or higher and standard library modules. No external dependencies are needed, making it straightforward to set up in your existing environment.
wannabeheroExtract and filter HTTP/HTTPS traffic from Charles Proxy .chlsj files.
PostHogAnalyze MCP-captured LLM agent traces from PostHog URLs or JSON exports.
andreichenchikProvide reusable React components for AI chat, code, audio, and persona UIs.
JingyueCongAnalyze PCAP files to identify communication issues and security anomalies.
