Skip to content

feat(browseros-agent): fetch LM Studio models live instead of a stale static list#1781

Open
GriggsRice wants to merge 1 commit into
browseros-ai:mainfrom
GriggsRice:feat/lmstudio-live-model-discovery
Open

feat(browseros-agent): fetch LM Studio models live instead of a stale static list#1781
GriggsRice wants to merge 1 commit into
browseros-ai:mainfrom
GriggsRice:feat/lmstudio-live-model-discovery

Conversation

@GriggsRice

Copy link
Copy Markdown

Summary

  • The LM Studio model picker sourced from a manually-run script snapshotting models.dev's generic registry (last refreshed weeks ago), never matching what's actually loaded on a user's machine. Picking a stale suggestion sent a model ID LM Studio didn't recognize, breaking chat and Test Connection.
  • Add useLMStudioProbe, a live query against LM Studio's native /api/v0/models endpoint (reports per-model load state + context length).
  • The picker now shows real local models, loaded one sorted first, with accurate context-length badges; the existing context-window auto-fill now uses real values.
  • Falls back to manual entry with a status hint when LM Studio is unreachable.

Test plan

  • bun run typecheck, bun run lint, bun run fallow — clean
  • bun test (app package) — 314 pass
  • Manually verified live against a running LM Studio instance via CDP: dropdown shows real loaded models with accurate context-length badges, loaded model sorted first, embedding models excluded

🤖 Generated with Claude Code

… static list

The Model picker for LM Studio providers sourced from a manually-run
script snapshotting models.dev's generic registry, last refreshed
weeks ago and never matching what's actually loaded on a user's
machine. Picking a stale suggestion sent a model ID LM Studio didn't
recognize, breaking chat and the Test Connection button.

Add a live probe (useLMStudioProbe) against LM Studio's native
/api/v0/models endpoint, which reports per-model load state and
context length. The picker now shows real local models sorted with
the loaded one first, and the existing context-window auto-fill
picks up accurate values instead of stale/missing ones. Falls back
to manual entry with a status hint when LM Studio is unreachable.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
@greptile-apps

greptile-apps Bot commented Jul 11, 2026

Copy link
Copy Markdown
Contributor

PR author is not in the allowed authors list.

@github-actions

github-actions Bot commented Jul 11, 2026

Copy link
Copy Markdown
Contributor

All contributors have signed the CLA. Thank you!
Posted by the CLA Assistant Lite bot.

@GriggsRice

Copy link
Copy Markdown
Author

I have read the CLA Document and I hereby sign the CLA

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.

1 participant