Agently

Sources

MCP Servers

Model Context Protocol (MCP) is an open standard that lets AI models safely communicate with external tools. Agently includes built-in MCP support for the most popular developer and productivity tools.

What is MCP?

MCP defines a standard interface between AI agents and tool servers. Each MCP server exposes a set of tools (actions the agent can call) and optionally resources(data the agent can read). The protocol handles authentication, serialization, and error reporting in a consistent way.

Because MCP is an open standard, the ecosystem is growing rapidly. Any MCP-compatible server can be added to Agently. See MCP in Agently for the full overview.

Built-in MCP Servers

Agently ships with pre-configured MCP servers for popular services. All you need is an API key or OAuth token.

GitHub

Read/write issues, PRs, code, and repos. Search across your org.

Linear

Create and update issues, manage cycles, read team workspaces.

Notion

Read and write pages, databases, and blocks in your Notion workspace.

Slack

Read channel history, send messages, search conversations.

Jira

Create issues, update status, search by JQL queries.

Figma

Read design files, components, and comments.

Connecting a Built-in Server

  1. 1.Open Settings → Sources → Add Source → MCP Server.
  2. 2.Select the service from the list (e.g. GitHub).
  3. 3.Follow the auth flow: for GitHub, you'll be redirected to authorize via OAuth. For others, paste an API token.
  4. 4.The server starts automatically. You'll see a green dot when it's connected.

Adding a Custom MCP Server

Any MCP-compatible server — including locally-run ones — can be added manually.

{
  "name": "My Custom Server",
  "transport": "stdio",
  "command": "npx",
  "args": ["-y", "@my-org/mcp-server"],
  "env": {
    "API_KEY": "your-key-here"
  }
}

For HTTP-based MCP servers, use "transport": "sse" and provide a url field instead.

Tool Approvals

MCP tools that write data (create, update, delete) require approval unless you have set the conversation permission to Auto.

When an agent wants to call a write tool, a confirmation dialog appears showing exactly what will be sent. You can approve, deny, or edit the parameters before the call executes.