Skip to content

feat: inject AI gateway model catalog into openclaw env#74

Merged
Iamlovingit merged 1 commit intoYuan-lab-LLM:mainfrom
Iamlovingit:multimodel
Apr 16, 2026
Merged

feat: inject AI gateway model catalog into openclaw env#74
Iamlovingit merged 1 commit intoYuan-lab-LLM:mainfrom
Iamlovingit:multimodel

Conversation

@Iamlovingit
Copy link
Copy Markdown
Collaborator

fix #68

Summary

  • Inject the full AI Gateway model catalog into OpenClaw instance env while keeping auto as the default gateway alias.

Changes

  • change CLAWMANAGER_LLM_MODEL from a single auto string to a JSON array like ["auto", ...active gateway models...]
  • keep OPENAI_MODEL=auto for compatibility
  • add backend tests for gateway model env injection
  • update instance env descriptions in the UI to reflect runtime-injected model catalog

Routing Behavior

  • sensitive-content governance is unchanged
  • even if a user selects a specific non-auto model, requests with sensitive content may still be rerouted to an active secure model by AI Gateway
  • if the selected model is already secure, it will continue to be used
  • if risk rules require blocking or approval, those rules still take precedence

@Iamlovingit
Copy link
Copy Markdown
Collaborator Author

/lgtm

@Iamlovingit Iamlovingit merged commit 0ffbc06 into Yuan-lab-LLM:main Apr 16, 2026
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

模型路由过于简短粗暴

1 participant