agent-proxy

Detect AI agent API protocols from a URL and route messages.

5|Updated Jan 3, 2026
One-click install
npx skills add https://github.com/yechao-zhang/red-team-agent-skills --skill agent-proxy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-proxy
Source: https://github.com/yechao-zhang/red-team-agent-skills/tree/main/.claude/skills/agent-proxy
Command: npx skills add https://github.com/yechao-zhang/red-team-agent-skills --skill agent-proxy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, aiohttp, websockets, websocket-client, gradio-client, and includes scripts (resource) components.

What problem does it solve?

Agent Proxy removes the manual work of figuring out how to talk to an AI service by automatically detecting the protocol and sending messages through the right interface.

Core Features & Use Cases

  • Auto-detection: Identifies OpenAI-compatible, Anthropic, Ollama, WebSocket, JSON, and Gradio endpoints from a single URL.
  • Conversation handling: Maintains message history, supports authentication, and returns the agent's replies in a consistent format.
  • Use case: Use it when you need to test an external AI agent, confirm its response format, or automate a back-and-forth conversation from Claude Code.

Quick Start

Give the target API URL to agent-proxy so it can detect the protocol and send a first message on your behalf.

Frequently Asked Questions about agent-proxy

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

FAQPage Schema
How do I automatically detect an AI agent API protocol from a URL?

To automatically detect an AI agent API protocol from a URL, the agent-proxy skill probes the endpoint to identify OpenAI-compatible, Anthropic, Ollama, WebSocket, JSON, or Gradio interfaces, then routes messages through the correct protocol automatically.

How do I automate conversations with an Ollama or Gradio endpoint?

You can automate conversations with an Ollama or Gradio endpoint by providing the target URL to the agent-proxy, which detects the specific protocol, manages message history, applies authentication, and returns replies in a consistent format.

Can I test external AI agents and log conversations from Claude Code?

Yes, you can test external AI agents and log conversations directly from Claude Code by passing the target API URL to the agent-proxy, which handles protocol routing, authentication, and conversation logging for automated testing workflows.

Does this approach support routing messages to WebSocket and OpenAI-compatible APIs?

Yes, this approach supports routing messages to WebSocket and OpenAI-compatible APIs, alongside Anthropic, Ollama, JSON, and Gradio endpoints, applying protocol probing and fallback handling to ensure message delivery across unknown interfaces.

What is the best way to handle fallback for unknown AI service interfaces?

The best way to handle fallback for unknown AI service interfaces is using the agent-proxy's protocol detection mechanism, which probes endpoints for specific protocols and applies fallback handling when an interface cannot be categorized.

Why do I need protocol probing for automated agent testing?

You need protocol probing for automated agent testing to eliminate the manual work of figuring out how to talk to a service, enabling automatic detection and routing of messages to OpenAI, Anthropic, Ollama, WebSocket, JSON, and Gradio endpoints.