Feature request: Proxy Recording Mode

[JuliusNmn]

I’d like to propose a simple “proxy recording” mode for MockLLM.

The idea is to allow MockLLM to forward incoming requests to a real inference provider, return the real response to the caller, and record the request/response pairs into a single YAML file.

Proposed Behavior

Configure MockLLM with:
• Upstream base URL
• API key

For each incoming request:
• Forward it to the real provider
• Return the real response
• Append all request and response pairs to a YAML file

This feature would make it much easier to generate realistic fixtures from actual inference without manually copying responses.

If there’s interest, I’m happy to prototype this and open a PR. Is this repo still active?