Support Custom API

[jgieringer]

Description

This PR serves as an example for developers bringing a custom endpoint.
This example also makes use of #104 where a conversation_id can be sent to the API and #107 where provider can start the conversation.
If the API rejects the initial conversation_id and returns its own, the client overwrites the initial with the API's conversation_id 🚀

[emily-vanark]

LGTM!

[Copilot]

Pull request overview

Adds a new “custom endpoint” LLM client implementation so developers can integrate an external chat API, including support for per-conversation conversation_id updates as introduced in PR #104.

Changes:

• Introduces EndpointLLM (aiohttp-based) and wires it into LLMFactory via endpoint-* model names.
• Adds endpoint configuration via environment variables and extends .env.example.
• Adds unit tests and fixtures for the new endpoint client.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
llm_clients/endpoint_llm.py	New custom endpoint LLM client using aiohttp + conversation_id metadata handling
llm_clients/config.py	Adds endpoint-related env vars and get_endpoint_config()
llm_clients/llm_factory.py	Allows creating EndpointLLM; updates judge-model error message
llm_clients/llm_interface.py	Tweaks conversation_id overwrite logic to ignore empty/same IDs
llm_clients/__init__.py	Documents support for endpoint / endpoint-*
.env.example	Adds example endpoint env vars
tests/unit/llm_clients/test_endpoint_llm.py	New unit tests for EndpointLLM
tests/unit/llm_clients/conftest.py	Adds mock_endpoint_config fixture

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.