Triage Agent

​

Overview

The Triage Agent automatically detects, investigates, and helps resolve production issues by connecting error monitoring, code repositories, team communication, and deployment tools.

​

Problem It Solves

When a production error occurs:

1. ❌ Someone has to notice (Sentry alert)
2. ❌ Investigate error (read stack trace, check logs)
3. ❌ Find relevant code (search GitHub)
4. ❌ Notify team (post to Slack)
5. ❌ Create ticket (open Linear issue)
6. ❌ Check if related to recent deploy (Render logs)

Total time: 20-40 minutes of manual work

​

Solution: Automated Triage

✅ Agent does all of this automatically in < 3 minutes

​

Required Connectors

​

Alternative Connectors

Don’t use these exact tools? Use alternatives:

​

Setup

​

Usage

Example prompts:

• “Check Sentry for new production errors and investigate”
• “Is the recent error spike related to our latest deployment?”
• “Investigate Sentry error SENTRY-ABC-123”

​

Key Features

• Error Detection: Monitor production errors from Sentry
• Code Investigation: Search GitHub for relevant code
• Team Notification: Alert team in Slack
• Deployment Tracking: Check recent Render deployments

​

Best Practices

• Filter Sentry Errors: Configure Sentry to only alert on actionable errors (exclude known/expected errors, filter by environment)
• Use Descriptive Commit Messages: Good commit messages help correlate errors with code changes
• Require Approval for Rollbacks: Keep destructive actions like rollback_deployment as “Require Approval” for safety

​

Next Steps