docs(providers): add the SCX.ai provider page - #808
Conversation
Documents the scx-ai route ahead of BerriAI/litellm#34752, which registers SCX.ai as a JSON-configured OpenAI-compatible provider Covers the two chat models on the route, GLM-5.2 and Qwen3.8 Max, with their context and output limits, plus SDK and proxy usage for streaming, function calling, structured output and image input. Every capability listed was checked against the live endpoint Also records two behaviours worth knowing: max_completion_tokens is sent upstream as max_tokens, and temperature is accepted in [0.0, 2.0) with 2.0 itself rejected, so LiteLLM clamps to 1.99 The page also gives SCX_API_KEY somewhere to live, which the env-key documentation gate in the main repo requires before that PR can go green
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Add SCX_API_KEY and SCX_API_BASE to the environment variables reference table so the credential names are discoverable from the proxy settings page, matching how the other providers list theirs.
|
This is the docs half of BerriAI/litellm#34752, and it is currently what gates that PR from going green. The code-quality and documentation jobs on #34752 check out litellm-docs main into docs/my-website, so test_env_keys.py cannot see the SCX.ai page until this merges, and it fails on SCX_API_KEY being undocumented. Latest push adds SCX_API_KEY and SCX_API_BASE to the environment variables reference table in docs/proxy/config_settings.md, so the credentials appear there alongside the other providers rather than only on the provider page. Verified locally by running tests/documentation_tests/test_env_keys.py from the litellm repo with docs/my-website pointed at this branch; it reports every environment variable documented, and the same run against main reproduces the failure. Could someone take a look when there is a chance. Happy to adjust anything on the page. |
|
@mubashir1osmani @ryan-crabbe-berri @tin-berri bumping this one, it is small and it is a hard gate on another PR. Three files, 238 additions, no deletions, a single provider page plus the sidebar and index entries. BerriAI/litellm#34752 cannot go green until this merges. I cannot request a reviewer from a fork, the API returns 404, so this comment is the only ping available. Happy to rebase or trim if anything here needs changing. |
|
bugbot run |
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
✅ Bugbot reviewed your changes and found no new issues!
Comment @cursor review or bugbot run to trigger another review on this PR
Reviewed by Cursor Bugbot for commit eb93d35. Configure here.
mateo-berri
left a comment
mateo-berri
left a comment
There was a problem hiding this comment.
LGTM. Thanks for the contribution!
The link pointed at a heading that doesn't exist; point it at the actual usage section instead. Pre-existing bug on main (added in BerriAI#808), unrelated to this PR, but it hard-fails every PR's Docusaurus build right now since onBrokenAnchors is set to throw.
2 participants
Adds the provider page for the
scx-airoute, which BerriAI/litellm#34752 registers as a JSON-configured OpenAI-compatible providerThe page covers the two chat models on the route,
scx-ai/GLM-5.2andscx-ai/Qwen3.8-Max, with their context and output limits, and shows SDK plus proxy usage for streaming, function calling, structured output and image input. Every capability listed was checked against the live endpoint rather than taken from a model listingIt also records two behaviours a caller would otherwise hit by surprise:
max_completion_tokensis sent upstream asmax_tokens, andtemperatureis accepted in[0.0, 2.0)with2.0itself rejected, so LiteLLM clamps to1.99This is also a prerequisite for the main PR. The
documentationandcode-qualityjobs there runtests/documentation_tests/test_env_keys.py, which checks this repo out and fails onSCX_API_KEYbeing read under./litellmbut mentioned nowhere in the docs. Verified locally that the page clears the gate:I work at SCX.ai
Note
Low Risk
Documentation-only: new provider page, sidebar entry, and env-var listings. No runtime or security logic.
Overview
Adds a provider page for the
scx-ai/route coveringscx-ai/GLM-5.2andscx-ai/Qwen3.8-Max, with SDK and proxy examples for streaming, tools, JSON schema, and vision.Documents
SCX_API_KEY/SCX_API_BASE(also listed in proxy env-var reference) and two upstream quirks:max_completion_tokensis sent asmax_tokens, and temperature is clamped to1.99because SCX rejects2.0.Links the page from the providers sidebar.
Reviewed by Cursor Bugbot for commit eb93d35. Bugbot is set up for automated code reviews on this repo. Configure here.