Capture to Sheet
Privacy policy
This notice describes the current development build’s local collection storage, user-triggered screenshot capture, and requests to your configured AI provider. Local storage does not mean every part of extraction stays on your computer.
Updated 2026-10-25 · In development
Back to the productWhat the extension saves
The extension uses Chrome’s local IndexedDB storage for projects, fields, rows, selected PNG screenshots, per-row photo crops, pending tasks, source URLs in full, and capture times. Successful rows also include extraction evidence and provider metadata such as model, token usage, and latency. This information supports collection, review, retries, and export.
The local service’s SQLite database separately saves project titles, instructions, columns, and revisions, along with queued task inputs and task results. Saved project definitions are used for later captures and survive a service restart.
From a visible tab to an AI request
Capture is initiated by you. Chrome first captures the full visible tab image temporarily so the extension can crop your chosen region locally. Only the selected region is persisted and submitted to the local service. This is not a scrolling screenshot or background crawler.
For extraction, the local service sends the selected PNG, task instructions, and field definitions to your configured provider. Large selections may also produce up to six bounded, lossless detail views of the same image, included in extraction requests. Optional numeric image-bound hints can accompany those requests; page HTML, DOM text, image URLs, cookies, and a separate source-URL field are not routinely sent. A URL or private information visibly present inside the selected pixels can still be read by the provider.
Local service configuration and secrets
Chrome local storage saves the pairing token and local service port used to connect the extension to your computer’s service, along with capture preferences such as the active project. The token is a local-service credential, not an AI API key.
Your AI API key is configured in the local service’s environment and stays out of the extension’s settings and this website. The local service uses that key to authenticate requests to the selected provider. Its database and configuration are local files under your control; do not share them in a support report.
The configured AI provider
Depending on your configuration, requests go to Google Gemini, OpenAI, DeepSeek, Alibaba/Qwen, or another compatible endpoint. That provider receives the submitted images, extraction instructions, and field definitions and handles them under the policies and terms applicable to your API account.
Some OpenAI-compatible request paths set store: false; this setting is not a universal flag across providers. Native Google Gemini requests use the GenerateContent API without an equivalent flag in this integration. No such setting establishes zero retention, excludes all provider logging, or replaces the provider’s policies. Provider requests can incur charges, including when a response is unusable, an optional multi-stage mode makes additional calls, or a retry is needed. Select only information you want that provider to receive.
Local retention and task cleanup
Chrome collections and screenshots have no automatic time-to-live and can remain until you delete them or the browser’s extension data is cleared. Exported files and device backups are independent copies.
The local service clears a task’s input field when it reaches a terminal state; tasks interrupted during processing are marked unknown and their input is cleared when the service starts again. Terminal task records older than seven days from their creation time are removed during service startup and when the processing queue drains. Cleanup is not a continuously running seven-day deletion timer, and saved project definitions are not covered by that expiry.
Website permission and capture controls
The extension connects to a local loopback service and asks for one optional broad website permission when you enable capture. This is a broad grant rather than a per-site authorization prompt. Actual capture is triggered by your selection on a supported normal webpage; protected browser pages and other unsupported surfaces can remain unavailable.
Capture OFF stops capture mode but does not revoke the website permission. Revoke website access through Chrome’s extension settings. Changing capture state or permissions does not delete saved collections, local-service data, exports, or copies held by the AI provider.
Deleting records and keeping copies
Deleting a project in the extension removes its Chrome records but does not delete the saved project definition from the local service’s SQLite database. Deleting one row from a multi-item capture preserves its sibling rows and shared screenshot; deleting the last row removes that shared capture from the extension.
Removing the extension, clearing extension storage, or losing a Chrome profile can delete local collections. The local service’s files, downloads, backups, SQLite deleted pages, and write-ahead log are separate. Deletion is not a promise of secure erasure. Export regularly and manage unwanted local-service files and independent copies separately.
Developer services, this website, and questions
The current product has no developer-operated cloud storage, analytics upload, or cloud sync. Extraction uses the service on your computer and the AI endpoint you configure. Local service startup logs do not routinely contain API keys, screenshots, or extracted values.
This website is hosted using OpenAI Sites and its infrastructure provider, Cloudflare. The hosting providers may process request URLs, IP addresses, and standard server logs to deliver and operate the website. Website analytics are currently not configured and hosting request processing is separate from the product’s collection and extraction requests. Use the shared support page for privacy questions, and remove sensitive content from any material you choose to share.
Contact and updates
This policy describes the current development build. It will be updated as the product’s data handling changes.