Agently

Sources

Sources Overview

Sources connect Agently to the world. They bring external data — from GitHub issues to your Obsidian notes — into the context window so your AI agents can reason over real information.

What are Sources?

A Source is a configured connection to an external data provider. When you include a Source in a conversation, Agently makes it available to the agent as a set of tools or read-only context.

Sources do not run automatically — you choose which Sources are active for each conversation, giving you precise control over what the agent can access.

Types of Sources

Most powerful

MCP Servers

Model Context Protocol integrations for GitHub, Linear, Notion, Slack, and more.

Flexible

REST APIs

Connect any HTTP endpoint with flexible authentication (Bearer, API key, OAuth).

Private

Local Filesystems

Grant access to local folders: Obsidian vaults, code repos, data directories.

Adding a Source

  1. 1.Open Settings → Sources (or press ⌘, then navigate to Sources).
  2. 2.Click Add Source and choose the type: MCP Server, REST API, or Local Folder.
  3. 3.Fill in the configuration (URL, auth credentials, or folder path).
  4. 4.Click Test Connection to verify it works.
  5. 5.Save. The Source is now available in your conversation sidebar under Active Sources.

Using Sources in Conversations

In any conversation, open the right panel and expand Sources. Toggle the Sources you want the agent to use. The agent will see them as available tools and call them automatically when relevant.

You can also enable a Source for all conversations in a workspace by default in workspace settings.

Permissions

Each Source respects the conversation's permission level:

  • Explore (read-only): The agent can read from Sources but cannot write or mutate data.
  • Ask to Edit: Write operations require your explicit approval per-action.
  • Auto: The agent can read and write freely. Use with caution.

See Permissions for full details.