Local Autocomplete Broken!

Before submitting your bug report

• I've tried using the "Ask AI" feature on the Continue docs site to see if the docs have an answer
• I'm not able to find a related conversation on GitHub discussions that reports the same bug
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS: mac
- Continue version: 
1.2.22
- IDE version:
vscode 
Version: 1.113.0
- config:
  
name: Local Config
version: 1.0.0
schema: v1
models:
  - name: qwen3.5-35b-a3b
    provider: lmstudio
    model: qwen/qwen3.5-35b-a3b
    apiBase: http://localhost:1234/v1/
    roles:
      - autocomplete
    requestOptions:
      timeout: 1200  # Set to 1200 seconds (20 minutes)
  
  OR link to agent in Continue hub:

Description

LM studio shows the following log every time i change code to trigger autocomplete. It was painful even getting this far with no docs on how to set up "The leading open-source AI code agent", only links to cloud services and telemetry options to disable.

2026-03-28 15:26:28  [INFO]
 [LM STUDIO SERVER] Running completion on text:  <fim_prefix># Path: soult...age_dir'])}",<fim_middle>
2026-03-28 15:26:28  [INFO]
 Streaming response..
2026-03-28 15:26:28  [INFO]
 [LM STUDIO SERVER] Processing...
2026-03-28 15:26:29  [INFO]
 [LM STUDIO SERVER] First token generated. Continuing to stream response..
2026-03-28 15:26:29  [INFO]
 [LM STUDIO SERVER] Client disconnected. Stopping generation... (If the model is busy processing the prompt, it will finish first.)
2026-03-28 15:26:29  [INFO]
 [LM STUDIO SERVER] Finished streaming response

I'm wondering if this is related to this? Your extension has been broken for nearly a year now?
https://github.com/continuedev/continue/issues/5079

I just want autocomplete, if anyone has a recommendation for an extension that does work, please let me know.

To reproduce

To reproduce, create proper documentation on setting up auto complete with local models from LM Studio (including on Mac). You'll see it doesn't work, and once you fix it, you can publish the missing documentation.

Log output

console.ts:139 [Extension Host] rejected promise not handled within 1 second: Error: Thank you for using GitHub Copilot. Your subscription has ended. 
leo	@	console.ts:139
ERR An unknown error occurred. Please consult the log for more details.
error @ log.ts:460
log.ts:460   ERR [Extension Host] Document /Users/os/.continue/config.yaml not found in AST tracker
error @ log.ts:460
console.ts:139 [Extension Host] Document /Users/os/.continue/config.yaml not found in AST tracker
leo @ console.ts:139
api.github.com/copilot_internal/v2/token:1  Failed to load resource: the server responded with a status of 403 ()
2console.ts:139 [Extension Host] deleteChain called from onDidChangeVisibleTextEditors
log.ts:460   ERR [Extension Host] Document untitled:Untitled-2 not found in AST tracker
error @ log.ts:460
console.ts:139 [Extension Host] Document untitled:Untitled-2 not found in AST tracker
leo @ console.ts:139
log.ts:460   ERR [Extension Host] Document untitled:Untitled-2 not found in AST tracker
error @ log.ts:460
console.ts:139 [Extension Host] Document untitled:Untitled-2 not found in AST tracker
leo @ console.ts:139
languageDetectionWebWorker.ts:241 ReferenceError: require is not defined
    at r.f.require (index.js:2:702403)
    at index.js:2:702071
    at Array.reduce (<anonymous>)
    at r.e (index.js:2:702049)
    at n.loadModel (index.js:2:705059)
    at n.runModel (index.js:2:705351)
    at n.detectLanguagesImpl (languageDetectionWebWorker.ts:239:41)
    at async u (languageDetectionWebWorker.ts:54:21)
    at async n.$detectLanguage (languageDetectionWebWorker.ts:81:19)
detectLanguagesImpl @ languageDetectionWebWorker.ts:241
api.github.com/copilot_internal/v2/token:1  Failed to load resource: the server responded with a status of 403 ()
console.ts:139 [Extension Host] deleteChain called from onDidChangeVisibleTextEditors
webviewElement.ts:491 An iframe which has both allow-scripts and allow-same-origin for its sandbox attribute can escape its sandboxing.
mountTo @ webviewElement.ts:491
api.github.com/copilot_internal/v2/token:1  Failed to load resource: the server responded with a status of 403 ()
console.ts:139 [Extension Host] deleteChain called from onDidChangeVisibleTextEditors
log.ts:460   ERR [Extension Host] Document untitled:Untitled-2 not found in AST tracker
error @ log.ts:460
console.ts:139 [Extension Host] Document untitled:Untitled-2 not found in AST tracker
leo @ console.ts:139
languageDetectionWebWorker.ts:241 ReferenceError: require is not defined
    at r.f.require (index.js:2:702403)
    at index.js:2:702071
    at Array.reduce (<anonymous>)
    at r.e (index.js:2:702049)
    at n.loadModel (index.js:2:705059)
    at n.runModel (index.js:2:705351)
    at n.detectLanguagesImpl (languageDetectionWebWorker.ts:239:41)
    at async u (languageDetectionWebWorker.ts:54:21)
    at async n.$detectLanguage (languageDetectionWebWorker.ts:81:19)
detectLanguagesImpl @ languageDetectionWebWorker.ts:241
log.ts:460   ERR [Extension Host] Document untitled:Untitled-2 not found in AST tracker
error @ log.ts:460
console.ts:139 [Extension Host] Document untitled:Untitled-2 not found in AST tracker
leo @ console.ts:139
api.github.com/copilot_internal/v2/token:1  Failed to load resource: the server responded with a status of 403 ()

Start by reproducing the reported configuration with Continue 1.2.22, VS Code on macOS, and the local LM Studio model, then compare the findings with issue #5079. Inspect the logged languageDetectionWebWorker.ts errors and autocomplete behavior. Done means the cause is identified, local autocomplete works or the limitation is documented, and setup instructions cover LM Studio on Mac.