# Build on AX Rush

> The scanner behind this product is callable: an anonymous HTTP API, an MCP server for agents, and machine-readable surfaces on every page. The public half needs no key and no account; an API key adds your own sites.

- [The HTTP API](https://axrush.com/developers/api): Three anonymous endpoints to measure any site, and an authenticated surface that reads your own with the panel's depth and says what to fix. OpenAPI included.
- [The MCP server](https://axrush.com/developers/mcp): Connect Codex, Claude Code, Cursor or Claude Desktop and scan from a conversation. Public reports, plus private reports, fixes and outcomes with an API key.

## Machine surfaces

Discovery is this product's own subject, so everything above is advertised the way we grade other sites on advertising it:

- [/.well-known/openapi.json](https://axrush.com/.well-known/openapi.json): the API, as a contract a machine can read
- [/.well-known/mcp/server-card.json](https://axrush.com/.well-known/mcp/server-card.json): the MCP server card
- [/.well-known/agent-card.json](https://axrush.com/.well-known/agent-card.json): the A2A agent card; the endpoint itself answers JSON-RPC at /a2a
- [/.well-known/ai-catalog.json](https://axrush.com/.well-known/ai-catalog.json): every agent surface this site serves, in one catalogue
- [/.well-known/agent-skills/index.json](https://axrush.com/.well-known/agent-skills/index.json): an Agent Skill teaching the scan procedure, with its digest
- [/llms.txt](https://axrush.com/llms.txt): the site index for language models

Every public page on this site also exists as Markdown: append .md to its path, or ask with an Accept: text/markdown header.

Scans announce themselves with a user agent that links to a page explaining what the crawler reads, how it honours robots.txt, and how to stop it. About the crawler: https://axrush.com/agent

---

Markdown-Darstellung von https://axrush.com/developers