Skip to main content
Agents can pull context from integrations to understand what to do. Context is the fuel — it gives agents enough detail to make good decisions.

​
What agents can read

IntegrationContext available
SentryError details, stack traces, breadcrumbs, affected users
LinearTicket description, comments, linked issues
GitHubCode, PR diffs, issue history, commit messages
PostHogSession replays, events, user properties
SlackChannel history, thread context

​
How context flows

When a Sentry alert triggers an automation:
  1. Trigger event includes error summary
  2. Agent can request full details: stack trace, breadcrumbs, tags
  3. Agent can also pull related context: recent commits, related Linear tickets
  4. All context is available in the agent’s prompt