Archyve Privacy Policy
Last Updated: August 26, 2026 · Architecture Version 2.0.0
This policy transparently explains how Archyve collects, processes, and stores data across our web application, API endpoints, and browser extension. We believe in architectural clarity: what is stored client-side vs. server-side, what external services are invoked, and how your AI keys are protected.
1. Bring Your Own Key (BYOK) & Key Storage
Archyve operates on a Bring Your Own Key (BYOK) model for AI synthesis. When you input your Google Gemini or Groq API key in Settings:
- Your API key is saved exclusively in your browser's
localStorage. - Your API keys are never stored in our database tables and are never written to permanent disk logs on our servers.
- When you generate a research dossier for a paper not yet in cache, your key is transmitted via secure HTTPS request headers (
x-ai-key) directly to our server route, used in-memory solely to authorize the completion request to your selected respective AI Provider, and discarded immediately after the stream closes.
2. Information We Collect
Depending on how you use Archyve, the following data points may be collected:
If you register an account, your account creds remain hashed. Unauthenticated users can still read existing public dossiers without creating an account.
When you submit a paper URL (e.g. IEEE, Springer, JSTOR, arXiv) for analysis, the URL and extracted public bibliographic identifiers are processed by our backend.
Generated paper summaries, reading assessments, and source citations are saved to our public papers index in our database to enable fast.
We log basic operational metrics including cache hit/miss status, trigger type (URL prepend vs extension), and pipeline latency in milliseconds to maintain performance.
3. External Service Providers & Data Flow
To construct validated research dossiers, Archyve communicates with specific external APIs:
Used for publisher web scraping to extract title, DOI, and author metadata from protected publisher web portals.
Public scientific metadata registries queried with paper titles and DOIs to verify bibliographic identity.
Queried with validated DOIs to discover legally accessible Open Access full-text PDF links.
Queried with paper titles to find publicly available code repositories associated with the research.
Receives structured metadata prompts using your provided API key to produce the research dossier.
4. Local Storage & Cookies
We utilize browser localStorage for user preferences and client-side credentials:
archyve_gemini_api_key/archyve_groq_api_key: Your local BYOK API keys.archyve_ai_provider: Selected provider (geminiorgroq).archyve_theme: Dark or light appearance preference.
5. Caching & Retention
Redis Caching: Paper dossiers are cached in Redis with a 7-day Time-To-Live (TTL) to deliver near-instant responses for subsequent queries.
Database: Public paper metadata (title, abstract, authors, DOI) and generated dossiers are stored permanently in the public corpus to allow all researchers to benefit from previously generated intelligence.
6. Your Rights & Contact
You have the right to delete your local credentials at any time by clearing your API keys in the Settings page or clearing your browser's site data.
For account inquiries, privacy questions, or data removal requests, please open an issue or reach out via our official GitHub repository: