Mobile Scan to Task

The mobile app ships a Scan to Task screen that turns a photo or PDF of a document into an editable draft task, mirroring the web app's /orgs/[orgId]/tools/scan-to-task scanner. Available from Org Tools → Scan to Task.

Flow

Pick a photo (camera or library) or a PDF → optionally add an instruction ("turn this into cleanup tasks") → tap Scan → review the generated draft(s), editing any field → Save to add it to the org's task list, or Discard to drop it. The task list refreshes automatically after a save.

Backend support

The mobile app calls REST endpoints added to the FriendChise web app under app/api/orgs/[orgId]/tools/scan-to-task/:

• upload-url (POST) — signed upload URL
• scan-to-task (POST) — runs the scan
• confirm (POST) — creates the task
• clear (POST) — discards a result

All reuse the existing requireOrgPermission(orgId, MANAGE_TASKS) bearer-auth guard, so no new auth mechanism was introduced. The scan/create logic lives in lib/services/scan-to-task-mobile.ts on the web app and writes to the same ScanTaskResult table the web scanner uses.

Permissions

Requesting the feature triggers the OS camera permission (NSCameraUsageDescription / Android camera permission) or photo library permission (NSPhotoLibraryUsageDescription) depending on the chosen source. Choosing a PDF uses the system document picker (expo-document-picker), which does not require an app permission prompt. Denied permissions show an inline, human-readable error instead of failing silently.

Picked photos are re-encoded to JPEG on-device (via expo-image-manipulator) before upload when they're HEIC/HEIF, since the backend's HEIC decoder (sharp/libheif) rejects some real-world iPhone photos — see Image Handling for why.

Testing on a physical device

1. Set EXPO_PUBLIC_API_URL to your computer's network IP (see Environment Variables) so the device can reach the backend.
2. Sign in, open an org, go to Org Tools → Scan to Task.
3. Take a photo of a document/checklist, or pick an existing photo/PDF.
4. Confirm the permission prompt, tap Scan, and wait for the review screen.
5. Edit a draft's fields and tap "Save to task list", then confirm it shows up on the Tasks screen.

Known limitations (v1, vs. the web scanner)

• One file per scan (web supports up to 12 files in a single batch).
• Only images (camera/library) and PDFs are supported as scan sources on mobile; the web scanner also accepts .docx, .txt, .md, .csv, and .json files.
• No merge/duplicate-adjudication UI — if a scanned draft looks like an existing task, the web app can suggest merging into it; on mobile, saving a task with an existing name can return an HTTP 409 conflict, and you must rename the task or retry after resolving it. The underlying AI duplicate-detection pass is also skipped for mobile scans to keep the request fast.
• No history/browsing screen — mobile only shows the results of the scan just run, not a persistent list of past scans (the web app's paginated history endpoint already exists and is a natural next step to add here).

TODO

• Add a mobile history/browsing screen backed by the existing paginated history endpoint.
• Run the AI duplicate-detection pass for mobile scans and surface a merge/duplicate-adjudication UI.
• Support multi-file batches and the additional file types the web scanner accepts.