What you need before you start
Microsoft states you must turn on generative orchestration to use MCP, so turn it on for your agent before you start (Microsoft Learn). You also need a KnowMan account, and, if you use API key authentication, an agent key created in the web app under Settings > Connections > Connect an AI agent (KnowMan MCP on GitHub).
Connect the server in the MCP onboarding wizard
In Copilot Studio, you connect an agent to an existing MCP server with the MCP onboarding wizard (Microsoft Learn). If you do not already have an MCP server, the documented integration flow includes creating one (Microsoft Learn). KnowMan provides a hosted remote MCP server at https://api.knowman.io/mcp using Streamable HTTP transport (KnowMan MCP on GitHub).
For authentication, the hosted server accepts OAuth, where the client opens a KnowMan consent page that you approve, or an API key sent as Authorization: Bearer <key> (KnowMan MCP on GitHub). The server address, transport, and authentication options are listed in the config snippet above.
Add the tools and resources to your agent
After connecting a server, you add its MCP tools and resources to your agent so the agent can use them (Microsoft Learn). A single MCP server can integrate and manage multiple tools and resources, and Copilot Studio agents can access each tool (Microsoft Learn).
The server provides the name, description, inputs, and outputs for its tools and resources (Microsoft Learn). Copilot Studio currently supports MCP tools and resources, and resources are file-like data an agent can read for more context, such as API responses or file contents (Microsoft Learn).
The six KnowMan tools
kb_search searches saved items by keyword or meaning and returns titles, URLs, and metadata. kb_ask answers a question with citations. kb_get_item returns the readable content of one item (KnowMan MCP on GitHub).
kb_save saves a URL to the Inbox or Knowledge Base with an optional note. kb_recent lists recent items. kb_memories lists saved claims, decisions, and playbooks (KnowMan MCP on GitHub).
Ask your agent to search or save, for example "search my KnowMan for ..." (KnowMan MCP on GitHub). See an MCP server knowledge base for how the pieces fit together.
Responsibility and optional publishing
When you connect to a non-Microsoft product, including an external MCP server, you are responsible for the tools and resources you access from within Copilot Studio (Microsoft Learn).
Optionally, you can publish the MCP connector so it can be used across tenants (Microsoft Learn). These MCP features apply to agents and agent flows powered by the standard harness (Microsoft Learn).
Updates and troubleshooting
When you update or remove tools and resources on the MCP server, Copilot Studio dynamically reflects these changes (Microsoft Learn). Microsoft also publishes a separate troubleshooting article for MCP integration issues (Microsoft Learn).
Questions people ask
Does Microsoft Copilot Studio support MCP servers?
Yes. You extend an agent with tools by using MCP, and Copilot Studio supports MCP tools and resources (Microsoft Learn). Each tool or resource a connected server publishes is automatically available for use in Copilot Studio (Microsoft Learn).
Do I need generative orchestration to use MCP in Copilot Studio?
Yes. Microsoft states you must turn on generative orchestration to use MCP (Microsoft Learn).
Which URL and authentication does the KnowMan MCP server use?
The hosted server runs at https://api.knowman.io/mcp over Streamable HTTP (KnowMan MCP on GitHub). It accepts OAuth sign-in or an API key sent as Authorization: Bearer (KnowMan MCP on GitHub).
How do I create a KnowMan API key?
In the KnowMan web app, open Settings, then Connections, then Connect an AI agent, and create an agent key (KnowMan MCP on GitHub). Keys look like km_..., and a key cannot create other keys (KnowMan MCP on GitHub).