ClaudeSwitch — AI skill for Claude Code
Native macOS utility for Claude Desktop 3P inference.
How to install ClaudeSwitch
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open xAmirHamza77/ClaudeSwitch and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What ClaudeSwitch does
Native macOS utility for Claude Desktop 3P inference. Features HTTP/1.1 persistent keep-alive proxy, wire-model rewriting, lowercase UUID fix, and one-click profile switching.
Alternatives in AI
- Inference Serving — ai-research-skills vLLM, SGLang, TensorRT-LLM, llama.cpp 5.4k ★
- Deepreasoning — A high-performance LLM inference API and Chat UI that integrates DeepSeek R1's CoT reasoning traces with Anthr 5.4k ★
- Vllm Mlx — High-performance OpenAI and Anthropic compatible LLM inference server for Apple Silicon 1.5k ★
README
ClaudeSwitch
**ClaudeSwitch** is a native macOS utility and menu suite designed for **Claude Desktop** (Anthropic 3rd-party inference interface, `deploymentMode: "3p"`). It enables you to seamlessly connect, switch between, and proxy custom LLM endpoints (Modal, Ollama, vLLM, LiteLLM, DeepSeek, AgentRouter, OmniRoute, and custom gateways) directly inside Claude Desktop.
Key Features
- Menu Bar Background Agent: Lives silently in your macOS top menu bar (
LSUIElement) without cluttering your Dock. Keep the proxy running in the background, switch profiles with one click, monitor proxy/Claude status, and restart Claude Desktop directly from the menu item. Closing the window keeps the proxy and background keep-alive active. - Any Gateway & Custom Endpoint: Connect Claude Desktop to local endpoints (Ollama, vLLM, LiteLLM) or remote providers (Modal, DeepSeek, AgentRouter, self-hosted gateways).
- HTTP Loopback Proxy & HTTPS Bypass: Claude Desktop strictly blocks remote non-loopback HTTP endpoints ("must use https or http on loopback"). ClaudeSwitch includes a built-in, lightweight loopback forwarder listening on
127.0.0.1with:- HTTP/1.1 Persistent Keep-Alive: Prevents socket drops and watchdog timeouts.
- Zero-Latency Unbuffered Streaming: Uses HTTP/1.1 Chunked Transfer Encoding to stream Server-Sent Events (SSE) tokens directly with zero buffering.
- Upstream Connection Pooling: Reuses warm TCP connections to remote gateways for sub-second responses.
- Wire Model Translation: Claude Desktop enforces Anthropic model naming patterns ("Doesn't look like an Anthropic model"). ClaudeSwitch lets Claude Desktop send standard names (like
claude-3-7-sonnet) while the proxy transparently translates requests to your upstream provider's model ID. - Schema Validation & UUID Fix: Prevents Claude Desktop's internal
Error: unknown config idand "Couldn't update saved configurations" by automatically generating and normal
Related Skills
AgentsMonitoring
Keep your AI coding agents (Claude Code, Codex…) alive in tmux and watch their status — dependency-free.
Brainrouter
LLM routing proxy for coding harnesses. Auto-routes to cloud or local inference via Bonsai LLM classification.
FuelSwitch AI
FuelSwitch AI — macOS menu bar app & floating HUD that monitors Claude Code, Codex and Gemini CLI usage limits
Agent Playbooks
Portable AI-agent playbooks installer -- fix bugs, build features, review code, and more, across any AI coding
Unfuck AI Comments
converts all comments inside go functions to lowercase
Neo Agent Brain
Point a cross-model AI engineering team at your own codebases. The Agent OS keeps named AI maintainers' identi
Related Agents
Small Business
COO and General Manager of the Small Business department, leading six employee skills that keep an owner-opera
UI Explorer
Drive and explore a native macOS app's GUI (screenshot → click/type → observe) and correlate each action with
Broca For Claude Desktop
Add persistent file-based memory to Claude Desktop in 5 minutes.