Add an opt-in setup guide for Strix's existing OpenAI-compatible endpoint settings, defaulting LLM_API_BASE to http://127.0.0.1:8000/v1 while allowing an environment override.
Keep LLM_API_KEY environment/secret-manager supplied and the model ID as an explicit placeholder to be replaced with the ID advertised by the FreeLLMAPI instance.
Link the guide from the LLM provider overview.
Validation
Environment settings load check passed for STRIX_LLM, LLM_API_BASE, and LLM_API_KEY without printing the test token.
tests/test_config_loader.py: 35 passed when excluding the POSIX 0600 permission assertion, which fails under Windows file-mode semantics.
### Issue 1
docs/llm-providers/freellmapi.mdx:31-33
A custom `LLM_API_BASE` does not make Strix choose chat completions. `resolve_api_type` chooses by model name, so a model such as `openai/gpt-5.4` still uses Responses. If the gateway only supports chat completions, requests will fail despite following this guide. Set `STRIX_API_TYPE=chat_completions` in the example, or explain when that override is needed.
```suggestion
- Strix chooses the API from the model's catalog entry, not `LLM_API_BASE`.
For a chat-completions gateway, set `STRIX_API_TYPE="chat_completions"`;
for a Responses gateway, set `STRIX_API_TYPE="responses"`. See
[advanced configuration](/advanced/configuration).
```
### Issue 2
docs/llm-providers/overview.mdx:64-66
The new guide is missing from the `LLM Providers` pages in `docs/docs.json`. Readers can find it through this overview card, but not through the provider sidebar. Add `llm-providers/freellmapi` alongside the other provider guides.
Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.
Summary
Adds a FreeLLMAPI setup guide using Strix's existing OpenAI-compatible settings.
Strix's provider guide explains how to connect a FreeLLMAPI gateway.
Preview approval required
This pull request comes from a fork, so Mintlify won't build a preview until someone with write access approves commit
39d6d8a.Approve preview