Connections
Give agents allow-listed access to external tools and services
Connections are secure integrations that act as the "hands" of your AI agents. They allow agents and Service Accounts to interact with external platforms safely on behalf of the team.
Communication Channels
Connect agents to messaging platforms like Telegram and Slack
Workflow Integrations
Connect agents to external services for automated workflows
MCP Servers
Connect agents to Model Context Protocol servers for extended capabilities
Tracing Servers
Connect observability and monitoring for agent execution
Security First
All connections are built on a strict security model:
- Allow-listing: Agents only access tools explicitly granted.
- Scoped Permissions: Granular control (e.g., "read issues" but not "delete repos").
- Team Boundaries: Connections respect Trust Boundaries and are scoped to the specific team they were created in.