InkWellOCR note capture

Scan · Read · Verify

Capture a note from photos

Add one or more photos of a note. InkWell reads the text, lets you verify it against the image, then merges every page into one continuous document you can edit and save.

What are you reading?
Drop photos here, or choose a source
JPG / PNG / HEIC · handwriting or print · paste with ⌘V works too

Saved per user · synced to agent

Your notes

A running list of everything you've captured. Notes are stored privately with the Vibes SDK and (optionally) mirrored so an MCP agent can read them.

Saved

Configure the AI · bring your own key

Settings

InkWell reads text with any OpenAI-compatible vision model. Pick a provider, paste a key, and set a model name. Calls are routed through the Vibes fetch proxy so cross-origin providers work.

AI provider

Model names are examples only — enter whatever your provider currently offers.

Stored per user with the Vibes SDK. Never shared with other users.

Recognition engine

Where text is read from your photos. The AI layer only re-formats the result afterwards.

Auto-format after captureRun the AI clean-up + YAML layer automatically once pages are combined into a note.

Behaviour

Route through Vibes proxyNeeded for most providers (CORS). Turn off only for CORS-enabled endpoints.
Mirror notes to MCP agentWrites each note to vibes.ai so an agent can pull it.
Auto-read images on addRuns OCR the moment a photo is added.
Auto-verify against imageSecond pass re-checks the reading; slower, more reliable.

Prompts

Tune how text is read and how it's standardised.

Blank = smart presets chosen by the document-type switch on the Capture tab.

Storage