Privacy Policy

• Account data: name, email address, and authentication data for people who sign up, are invited to, or administer a workspace. Sign-in is primarily magic-link email; we also support existing password-based accounts.
• Workspace configuration: subdomain, brand colors, logo, site settings, published package metadata, role assignments, publishing credential metadata (token names and creation times; secrets are shown once and not stored in recoverable form), and source connection metadata.
• OAuth connection data: when a workspace connects a third-party source (for example GitHub), we store the tokens and identifiers needed to sync that source, encrypted at rest, for as long as the connection remains enabled.
• Contact form submissions: the name, email address, and message you provide when you contact sales through the marketing site, plus anti-abuse checks on that form.
• Transactional email: magic links, email verification, and invitations sent through our email service so you can sign in and join workspaces.
• Security and audit events: administrative actions (invites, role changes, membership changes), successful and failed sign-in events when recorded, and related timestamps, used to operate and secure the Service.
• Technical data on the apex: IP addresses and related request metadata used for rate limiting, abuse prevention, and securing authentication. This is separate from workspace analytics, which do not store raw IP addresses.
• Usage and credits: per-workspace meters, balances, and ledger entries for metered features (including AI and index work), so limits and billing can be enforced accurately.
• Billing references: when you buy a plan or credit pack through self-serve checkout, we store Stripe customer, subscription, and payment references needed to grant entitlements and credits. Payment card details are entered on Stripe-hosted pages and are not stored by DevDocs.
• Operational reliability data: application error events and performance traces for the hosted.devdocs.ai Worker and apex app (for example request path, stack trace, and related diagnostic metadata), and platform operational logs drained for incident response. We configure error monitoring to avoid sending request bodies and default personally identifiable information where the product allows it.
• Marketing site interaction: optional aggregate analytics when enabled and consented (see Cookie Policy). The apex marketing FAQ assistant answers only from our FAQ database and does not send your question to a language model.

For each workspace, on behalf of its operator, we process:

• Customer Content: published documentation packages, and content Customer authorizes through connected sources, so we can host, index, embed for retrieval, and serve that workspace.
• Aggregate site analytics: country, city or network attributes from the edge request, referring host, and top paths, computed without storing raw IP addresses. We do not compute or store a cross-workspace visitor identifier; one workspace's numbers cannot be correlated with another's through our analytics design.
• AI assistant questions: the free-text question a visitor submits, processed to generate a grounded answer (and sent to our model provider for that purpose when a model call is required), plus the answer and citations returned.
• Content gaps and feedback: unanswered or low-quality questions and thumbs feedback retained so the workspace operator can improve documentation. Feedback is not used to train a model.
• Usage metering: counts of AI assistant use and related metered work per workspace. Metering records do not store full question text.
• Client-side chat history: multi-session chat history for the reader widget is stored in the visitor's browser local storage on that workspace host only. It is not a server-side conversation archive operated by DevDocs for our own use.

DevDocs does not train, fine-tune, or otherwise improve any machine learning model on workspace documentation content or visitor questions, and does not permit its subprocessors to do so. This applies equally to a workspace's published documentation, connected source material, and the questions its visitors ask, and it is stated the same way in our Terms of Service, our Data Processing Agreement, and on our security page.

Feedback submitted through the assistant's thumbs up or down control is used only to help a workspace operator see which of their own questions went poorly answered. It is never sent anywhere that trains a model.

We use a small number of subprocessors, listed in full at hosted.devdocs.ai/legal/subprocessors. That list is the source of truth for vendor names and purposes. In short: Cloudflare runs the platform; Anthropic is the language model behind workspace answers when a model call is required (reached only through Cloudflare AI Gateway; embeddings use Cloudflare Workers AI); Better Stack receives operational logs via Cloudflare Logpush for reliability and incident response; Sentry receives application error monitoring and performance traces for the Worker and apex app; Stripe processes self-serve plan and credit-pack payments.

When a workspace question requires a model call, the question and retrieved documentation chunks are sent to Anthropic through Cloudflare AI Gateway. Anthropic's commercial terms for this use prohibit training on the content we send. Payment card data is handled on Stripe-hosted Checkout pages, not on our forms. Workspace documentation hosts do not load Sentry or Better Stack as marketing trackers.

We do not currently make a data residency commitment (a promise that data is stored only in a specific geography). Our governing law and venue clause in the Terms of Service is a choice of forum, not a residency guarantee, and we say that plainly rather than implying one.

hosted.devdocs.ai (the apex, where we are the controller) uses a session cookie for sign in, a consent preference cookie, and local theme preference storage. Non-essential marketing analytics, when enabled, require consent. Workspace documentation hosts do not set non-essential HTTP cookies and do not show a DevDocs cookie banner. They may use first-party local storage for theme preference and reader chat history on that host only. See the Cookie Policy for detail.

Account and workspace data is retained for the life of the workspace and deleted within a commercially reasonable period after termination, as described in the Terms of Service. Usage metering and credit ledger records are retained longer because they are the historical basis for limits and billing and cannot be reconstructed once lost. Aggregate workspace analytics are retained as rolling operational data for a limited window, not indefinitely. Content gap and feedback signals are retained so the workspace operator can improve documentation. Client-side chat history remains until the visitor clears site data for that host. On termination, Customer Content is removed from active serving and primary stores as described in the Terms and the DPA.

Depending on your location, you may have rights to access, correct, delete, or export your personal data, or to object to certain processing. If you are a workspace visitor, please direct requests to the workspace operator first, since they are the controller of your data on their site. For account data where DevDocs is the controller, contact privacy@devdocs.ai.

Questions about this policy can be sent to privacy@devdocs.ai.