Skip to main content
Actions are operations that agents perform on external services. Beyond writing code and opening PRs, agents can create issues, post to Slack, update tickets, resolve errors, and call any tool exposed by an MCP server. Actions respect your permission settings. Configure what is allowed, requires approval, or denied in the Permissions page.
Actions configuration showing available integrations and their permissions.

​
Available actions

When you connect an integration, its actions become available to agents automatically. Each integration exposes a set of read, write, and danger-level operations that flow through the permission system.

GitHub

Create PRs, push commits, comment on issues, merge branches, and more.

Linear

Create, update, and comment on issues. List and filter across teams and projects.

Sentry

List and inspect issues and events. Resolve, ignore, or reassign errors.

Slack

Post messages to channels and threads.
Any MCP server you connect also surfaces its tools as actions. These go through the same permission and approval system.

Add an MCP Server

Connect any MCP-compatible tool- Neon, Stripe, Playwright, Context7, and more.

​
Setting up actions

Actions are enabled by connecting integrations. No separate configuration is needed- once an integration is connected, its actions are immediately available to agents.
1

Connect an integration

Go to Settings > Integrations and connect the services you want agents to use (GitHub, Linear, Sentry, Slack, or any MCP server).
2

Verify action availability

Once connected, the integration’s tools appear as available actions for agents to use.
3

Configure permissions

Set allow, require approval, or deny in Permissions, including automation-level overrides when needed.
For the full permission model, risk levels, and approval behavior, see Permissions.

​
Next steps

Permissions

Configure which actions are allowed, denied, or require approval.

Add an MCP Server

Connect external tools and services via Model Context Protocol.