Extending LocalLens
Three concrete next steps, each grounded in a QVAC capability you already have.
LocalLens is finished as a sample. To turn it into something that fits your workflow, you'll want to extend it. The most natural extensions reuse capabilities the QVAC SDK already exposes: OCR for documents, speech-to-text for voice, and the existing model state for richer diagnostics.
The pattern is the same every time:
- Find the file that owns the responsibility.
- Make the smallest change that fits the existing seams.
- Keep extension-specific logic out of modules that don't need it.
Three places to start
PDF and image parsing with QVAC OCR
Use @qvac/sdk's ocr() to turn images and scanned pages into LocalDocuments — the rest of the pipeline doesn't change.
Voice questions with QVAC transcription
Record audio in the browser, transcribe locally with Whisper through @qvac/sdk, send the text to /api/brains/:id/chat.
Better diagnostics
Surface model loading, indexing progress, file/chunk counts, active model — without giving the UI the ability to mutate state.