Start

Connectors

Connectors

Connectors translate source-specific resources into AgentGraph entities, people, edges, observations, and searchable content.

AgentGraph's bundled connectors are integrations for common services, and examples of an open connector pattern. Each connector owns its URLs, authentication, fetch logic, refresh behavior, and source metadata while returning the same graph-shaped output to core AgentGraph.

ConnectorWhat it lets the agent perceive
GmailEmail threads, participants, subjects, bodies, and attachment references
Google DriveFolders and files, content, ownership, and containment
Google DocsDocument content, ownership, authorship, and source dates
Google SheetsSheet values, ownership, authorship, and source dates
SlackChannels and DMs, messages, replies, authors, mentions, and attachments
DiscordChannels, DMs, threads, messages, authors, mentions, and attachments
RSSFeeds, posts, dates, authors, and publication relationships
WebConfigured pages and bookmarks with titles, text, metadata, and URLs

#Extensible by design

A connector is a Python package that translates a service into local entities, people, edges, observations, and searchable content. Connectors can be built around:

  • an API, export, or webhook;
  • a local database or structured file;
  • a browser-accessible resource;
  • an internal system used by one team; or
  • a niche public service maintained by its own community.
Bring any service into your agent's world. Build a private connector for internal context, publish an integration for a tool you use, or contribute a connector that provides an integration for others to use.

See Extending AgentGraph for the complete interface, example connector, packaging, and registration instructions.

#Community direction

The bundled connectors prove the architecture; they are not presented as a complete integration ecosystem. The long-term direction is a community-maintained library spanning developer tools, issue trackers, notes, calendars, research libraries, media, CRMs, support systems, and the unusual services that matter to individual users.