Harness Bridge
SDK / ToolA local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.
- launcher
- local-models
- cli
- menu-bar
AI gateway with a Rust core — call Claude and 100+ providers through one OpenAI-format API.
LiteLLM is an AI gateway and SDK (Rust core, Python SDK) that fronts Anthropic, OpenAI, Bedrock, Vertex, and 100+ providers behind one API surface, with cost tracking, budgets, rate-limit handling, fallbacks, and load balancing. Route Claude traffic through it when you need spend controls or provider failover; the proxy ships virtual keys and per-team budgets for production deployments.
When your agent code should survive a provider swap or outage — LiteLLM normalizes the API, tracks cost per key, and can route MCP tools through the same gateway.
pip install litellm
MCP gateway client config:
{
"LiteLLM": {
"url": "http://localhost:4000/mcp/",
"headers": {
"x-litellm-api-key": "Bearer <your-master-key>"
}
}
}OAuth for upstream MCP servers can hit providers that refuse dynamic client registration — pre-register the OAuth app and set credentials.client_id/secret. The proxy is a real service to operate (keys, spend limits), not a library call.
Repository
BerriAI/litellmCharted
A local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.
Answers macOS Spotlight and Siri prompts with your signed-in Claude Code account — an experimental macOS 27 model-delegation provider.
Extract your own local chat history from ten AI coding assistants into one normalized JSONL — for fine-tuning, analytics, or backing up years of conversations. Standard library only.