Skip to main content
The ReclaimLLM local API proxy (rclm-proxy) sits between your application and LLM providers. Every request and response passes through it, gets recorded, and is uploaded to your dashboard asynchronously — without ever blocking your requests. Setup requires no changes to your application code: you set one environment variable pointing at the proxy, and your existing provider API keys stay exactly where they are.
This is not the Enterprise LLM Gateway. The local proxy is rclm-proxy on http://localhost:9800, uses your provider API keys, and is for individual capture. The hosted gateway is https://proxy.reclaimllm.com/<org-slug>/v1, uses a team gateway key (rclm-gw-…), and is enterprise-only. Do not point SDKs at proxy.reclaimllm.com when you mean rclm-proxy, or the reverse. See Enterprise Gateway.
The local API proxy is experimental. You may encounter rough edges. If you run into issues, please report them so the team can improve it.

​
Supported providers

  • Anthropic
  • OpenAI
  • Azure OpenAI
  • Gemini (via LiteLLM)

​
Set up the proxy

1

Install the package

Install rclm with optional proxy dependencies:
The rclm[proxy] variant includes LiteLLM, which enables support for all provider backends. Use this if you plan to proxy Gemini or Azure OpenAI traffic.
2

Set your RCLM API key

Export your ReclaimLLM API key as an environment variable:
Your existing provider API keys (Anthropic, OpenAI, etc.) stay in place — the proxy forwards them transparently to the upstream provider.
3

Start the proxy

Start the proxy server:
Expected output:
The proxy listens on port 9800 by default.
4

Point your application at the proxy

Update your application’s base URL environment variable to route traffic through the proxy. Your provider API keys remain unchanged.
Your existing provider API keys stay in place — the proxy forwards them to the upstream provider transparently. You do not need to change or share those keys with ReclaimLLM.

​
Verify it’s working

After starting the proxy and updating your environment variables:
  1. Make a request from your application as you normally would.
  2. Open your RCLM dashboard — the session should appear within a few seconds.
  3. Click into the session to inspect the request, response, model, and token counts.
You can also confirm the proxy is running by checking whether port 9800 is active:

​
How the proxy handles your traffic

  • Async upload — session data is uploaded to ReclaimLLM in the background after the response is returned. Your requests are never delayed waiting for the upload to complete.
  • Runs locally — the proxy process runs on your own machine. Traffic between your app and the proxy never leaves your system until it reaches the upstream provider.
  • No code changes — only environment variables change. Your application calls the same SDK methods and passes the same provider API keys.
  • LiteLLM-powered routing — when installed with rclm[proxy], the proxy uses LiteLLM to normalize requests across providers so all supported backends work through the same local endpoint.
Add the RECLAIMLLM_API_KEY and *_BASE_URL exports to your shell profile or .env file so the proxy configuration persists across sessions without re-exporting manually.