feat(settings): probe the active provider, calm idle-service tests, context length for managed LM Studio - #157
Open
nixfred wants to merge 1 commit into
Open
Conversation
…ontext length for managed LM Studio WHAT: three Settings-view fixes that make the panel tell the truth about the running configuration. 1. The LLM Test button and model-list caption now target the endpoint the pipeline actually uses: managed lm-studio → :1234, managed ollama → :11434, external → the lmStudioUrl field. Previously a healthy ollama-managed setup showed a red 'connection refused' test against LM Studio's port. 2. 'Connection refused' on a lazily-spawned service (whisper-server always; the LLM in managed modes) now renders as a neutral '○ not running — starts on demand when needed' instead of a red error. Genuine problems (wrong port answering 404, timeouts) stay red. 3. New llmContextLength setting (default 0 = LM Studio's model default) passed as --context-length when the managed provider auto-loads the model. Without it, respawned models come up with the model's saved default context (often 4k) and silently truncate long transcripts. Exposed as a dropdown only when the LM Studio provider is active. TESTED: supervisor tests extended (context pass-through + default-0 call), settings-repo tests green, eslint + tsc clean on both configs. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Fx2epy4XwaBS3S8Y5J7FX9
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three Settings-view fixes that make the panel tell the truth about the running configuration. Found while running an ollama-managed install where every Test button was red on a perfectly healthy stack.
Provider-aware Test + captions. The LLM Test button and the model-list caption now target the endpoint the pipeline actually uses (managed lm-studio →
:1234, managed ollama →:11434, external → thelmStudioUrlfield), mirroring the LMStudioClient wiring in main. Previously an ollama-managed setup showed✗ connection refusedagainst LM Studio's port while summaries worked fine.Idle ≠ broken.
connection refusedon a service the app lazily spawns (whisper-server always; the LLM in managed modes) now renders as○ not running — starts on demand when neededinstead of a red error. Genuinely wrong states (another service answering 404 on the port, timeouts) stay red.Context length for managed LM Studio. New
llmContextLengthsetting (default 0 = model default, preserving current behavior) passed as--context-lengthwhen the supervisor auto-loads the model. Without it a respawned model comes up with its saved default context — often 4k — and silently truncates long transcripts. Exposed as a dropdown only when the LM Studio provider is active, with a warning about the RAM tradeoff.Testing: supervisor tests extended (pass-through + default-0), settings-repo suite green, eslint + both tsconfigs clean, verified live on macOS 26 / M4 Max with both ollama and lm-studio providers.
🤖 Generated with Claude Code
https://claude.ai/code/session_01Fx2epy4XwaBS3S8Y5J7FX9