Docs site privacy

This page explains what the public AxiomCore websites measure and store: the landing site (axiomcore.dev), this documentation (docs.axiomcore.dev), and the Playground (playground.axiomcore.dev). Reading the docs needs no account, contract, project, or CLI installation.

Analytics at a glance

Cloudflare Web Analytics is separate from the optional choice. See Cloudflare's Web Analytics privacy explanation.

Clarity never loads in the Playground, where you can edit source code and preview your own content. Session replay is limited to the landing site and the docs.

Your choice

• The banner offers Decline and Accept analytics. Declining loads neither Google Analytics nor Clarity.
• Privacy choices reopens the banner so you can change your decision.
• The choice is stored in a first-party cookie (axiom_analytics_consent) on axiomcore.dev, shared by the three sites, for six months.
• If you decline after accepting, the sites deny analytics consent, delete the Google Analytics and Clarity cookies, and reload the page.
• Accepting allows usage measurement only. This integration disables advertising storage, Google signals, and ad personalization.

What Google Analytics receives

• The page origin and path.
• The utm_source, utm_medium, and utm_campaign parameters, only when each is a short, simple campaign label (up to 80 letters, digits, ., _, ~, or -).
• The origin of the referring site, not its full URL.
• Custom events from a fixed list of action categories, such as opening the documentation, copying an installation command, running an example, or selecting an Inspector view. An export event records that a download started, not a confirmed local install.

Other query parameters and URL fragments are not sent to Google Analytics. Source code, request bodies, generated files, raw search queries, and compilation error messages are never event data. Clarity may process the page URLs and referring URLs of pages where it runs, so do not put private data in public-site URLs.

Provider retention, regional processing, and access settings are managed in the hosting and analytics accounts. No fixed retention period is promised here. Send questions or requests to contact@yashmakan.com.

Search and machine-readable routes

Documentation search uses the site's generated public content. The site also serves public processed MDX, llms.txt, llms-full.txt, a sitemap, robots rules, Open Graph images, a version file, and a health route. Do not put secrets, private URLs, customer data, or unreleased implementation details in public MDX or metadata.

Ask AI

Ask AI is an optional chat on the documentation site. It runs only when you send a question, and it is not covered by the analytics choice above.

• When you send a question, the site sends the conversation in that chat (your messages, the earlier answers, and the title and path of the page you have open, without its query string) to OpenRouter, which passes it to the AI model chosen for the docs. OpenRouter and the model provider handle it under their own terms and privacy policies.
• The documentation site does not store conversations. The chat lives in your browser tab and is gone when you reload the page or start a new chat. Only an unsent draft is kept in your browser's local storage, so it survives a reload.
• Opening Ask AI loads Cloudflare Turnstile, which checks that a person, not an automated script, is asking. Most visitors see nothing; some are asked to select a checkbox. Cloudflare processes browser and device signals for this check under the Turnstile privacy addendum. Turnstile does not load until you open Ask AI.
• To limit abuse, each visitor can ask a limited number of questions per minute, hour, and day, and the site has a daily total. Questions are counted against a keyed one-way hash of your network address, never the address itself, and each count is deleted when its period ends, at most one day later.
• Answers come from the public documentation and can contain mistakes. Do not put secrets, credentials, private code, or personal data in a question.

Hosting logs

The hosting platform or network edge may create operational request logs. The documentation repository does not define their retention, regional storage, or access policy; those controls belong to the deployment environment and its owner. The health endpoint returns only a schema version, a status, the service name, and an optional public source revision.

Feedback and diagnostics

Anything you submit through a repository, community, or private security channel follows that channel's controls. Sanitize diagnostic output as described in Support and feedback. Never send access tokens, cookies, signing material, DSNs, sandbox keys, private contracts, or sensitive request or response bodies.

Separate telemetry surfaces

Website analytics is separate from the CLI and from your application:

Installing or running the CLI does not depend on visiting these sites, and viewing these sites does not configure runtime telemetry.