Docs MCP server
The public, unauthenticated MCP server that lists, searches, and reads these documentation pages.
The Vidext Documentation MCP server gives any MCP client read access to these docs. It needs no account and no token, so an agent can learn how Vidext works before anyone signs in.
https://app.vidext.com/api/mcp/docsIt reports itself as vidext-docs (title Vidext Documentation). It never changes anything; to read an organization's own knowledge and modules, use the Operator server.
Connect
claude mcp add --transport http vidext-docs https://app.vidext.com/api/mcp/docsAny client that speaks MCP 2026-07-28 over streamable HTTP can connect the same way, with no authentication step. The transport rules are the Operator server's: POST with the _meta envelope and the Mcp-Method header. See transport details.
Tools
| Tool | Arguments | Returns |
|---|---|---|
list_docs | section (optional: platform, learn, or api), locale (optional: en or es) | count and pages, each with path, title, description, and url. |
search_docs | query (required, 1–200 characters; keywords work better than a full question), locale | query, count, and up to 10 pages, each with path, title, url, and up to 5 matching passages. |
read_doc | path (required, for example /docs/api/mcp/authentication), locale | The page as markdown. Pages longer than 40,000 characters are cut, with a link to the full page. |
Paths look like /docs/platform/quickstart. Pass locale: "es" for Spanish; a page without a Spanish version comes back in English.
All three tools are marked read-only. A failure (an unknown path, the docs site not answering) comes back as a result with isError: true and a plain-text message. For an unknown path, call list_docs for valid ones.
Resource
| Resource | URI | What it is |
|---|---|---|
| Docs index | vidext-docs://llms.txt | The docs llms.txt: one line per page with its title, path, and description. |
What it reads
The server reads the public docs site at https://docs.vidext.io: its llms.txt index (cached for up to ten minutes), its search, and the markdown version of each page. It sees exactly what a visitor to the docs sees, nothing from any organization.
Last updated on