Effective October 2, 2026. Published by thedemiurge (GitHub: scottyad).
When you use Citation Capture, it reads information from the webpage you are capturing, including its URL, title, author and publication metadata, DOI or arXiv identifier, available abstract or page text, and selected text. This information is used to prepare a citation and literature note. The extension does not need your complete browsing history.
To enrich citations, the extension can send paper identifiers or search information derived from the citation to services such as DOI.org, Crossref, arXiv, and PubMed. Those services receive the request and associated network information, such as an IP address.
Cloud AI can run automatically when you open the capture popup, depending on your AI settings. In cloud mode, text used for summarization and titles or abstracts used for tag suggestions are sent to the configured backend, normally cite.archpanda.xyz. The backend sends the relevant text to Ollama Cloud for processing. This can happen before you click Capture to Obsidian.
In version 1.0.7 and later, every enabled AI mode uses the hosted ArchPanda backend and its server-side Ollama Cloud account. Personal provider keys and custom AI endpoints are not used. AI can be disabled in Settings. Older versions may still use their previously selected provider; update to version 1.0.7 to use this routing. Review the AI mode in Settings before capturing confidential material.
Ollama Cloud processes hosted AI requests under its own terms and privacy policy. Cloud AI does not automatically fall back to a different provider or to local Ollama.
The extension stores settings in browser extension storage, including vault and folder names, note templates, AI and bridge preferences, configured API keys or local REST tokens, license information, and usage counters. Credentials are not published as part of your note. When you save a note, its content is passed to Obsidian using your selected method: an Obsidian URI, the companion plugin, or a configured local REST endpoint. Copying a citation places it on your clipboard. Obsidian vault storage, sync services, clipboard tools, and configured endpoints have their own data practices.
License keys are sent to the backend for verification and access to paid features. The service maintains license, tier, and credit or usage information. Purchases are processed by Stripe. Payment-related records may include customer email, checkout identifiers, purchased tier, and issued license details. Payment card information is entered through Stripe checkout rather than the extension's capture form. See Stripe's privacy policy.
Network services may produce operational logs containing IP addresses, request paths, timestamps, and errors. Citation content is transmitted for metadata resolution and AI processing; this policy does not promise zero retention by infrastructure or AI providers. License and transaction records may remain after you uninstall the extension for service administration and applicable recordkeeping requirements. Browser settings remain until cleared or the extension is removed. Notes already saved in Obsidian remain until you delete them.
You can change AI providers and modes, remove configured credentials, use citation copying, and delete notes from your vault. Uninstalling the extension removes its browser extension storage but does not delete Obsidian notes, server license records, or provider records. Do not send private or sensitive webpage content to a cloud or remote service unless you intend that service to process it.
For support or a request about service-held data, contact the maintainer through the project's GitHub issues. Do not post API keys, license keys, payment information, or other sensitive data in a public issue; request a private contact method instead. Changes to this policy will be published here with an updated effective date.