MediaKind Context MCP | MediaKind Docs

MediaKind Context MCP

Give your AI tool direct access to published MediaKind documentation and API references. The MediaKind Context MCP is a read-only server hosted on this documentation site. It works over Streamable HTTP and does not require authentication.

Connect

Point your MCP client at:

https://docs.mediakind.com/api/mcp

No API keys, tokens, or headers are needed.

If your client accepts a JSON config:

{

"transport": "streamable-http",

"url": "https://docs.mediakind.com/api/mcp"

}

Claude

Add a custom connector in Claude and paste https://docs.mediakind.com/api/mcp as the remote MCP server URL. Leave authentication fields empty.

Claude Code, Codex, and other agents

Add a server named mediakind-context with URL https://docs.mediakind.com/api/mcp. No headers or authentication values are needed.

What you can search

The server covers the full published MediaKind documentation corpus:

Content Description
MK.IO docs Guides, how-tos, and concepts for MK.IO
Beam docs Guides, how-tos, and concepts for MK.IO Beam
Media API OpenAPI reference for the Media API
Management API OpenAPI reference for the Management API
Fleets API OpenAPI reference for the Fleets API
Infrastructure API OpenAPI reference for the Infrastructure API
Release notes Changelogs and feature roadmap entries

Tools

The server exposes seven read-only tools. None of them generate answers — they return documentation content that your AI tool uses to compose its response.

search

Run a semantic search across the entire documentation corpus. Returns the most relevant chunks ranked by similarity.

You can optionally pass source_groups to narrow results to specific content areas — for example, only MK.IO docs or only the Media API.

list_sources

Browse all available documentation sources. Each source represents a page or API specification. Useful for discovering what content is available before diving deeper.

Supports optional source_groups filtering and a query substring filter.

list_sections

List the individual sections within a specific source. Use this after list_sources to see the structure of a page before reading it.

get_doc_section

Retrieve the full content of a specific section by its source URL and section index. Use this to read a particular part of a documentation page.

get_endpoint

Look up a specific API endpoint by its signature (e.g. GET /api/v1/assets). Returns the full endpoint documentation including parameters, request body, and response schemas.

list_context_groups

List all the content groups available on this server, with their names and descriptions. Useful for discovering what source_groups values you can filter by.

server_info

Returns server metadata including the protocol version, server name, and which content groups are active.

Filtering by content area

Most tools accept an optional source_groups parameter. Pass one or more group keys to narrow results to specific parts of the corpus:

Omit source_groups to search everything.

How it works

The MCP server does not generate answers. When your AI tool calls a tool, the server retrieves the relevant documentation chunks from the published corpus and returns them as structured content. Your AI tool then uses that content to write its response.

All tools are read-only and idempotent. The server does not modify any data and repeated calls with the same arguments return the same results.