ErrorTriage

Overview

ErrorTriage is an AI-powered error monitoring and incident management agent that automatically detects, diagnoses, and tracks critical issues across your stack. It monitors Sentry for high-severity errors, generates intelligent root-cause analysis using advanced LLMs, creates structured Linear tickets with full context, and notifies engineering teams via Slack—all while preventing duplicate ticket creation through intelligent deduplication.

How It Works

1. List Issues (Sentry) — Continuously scans Sentry for new unresolved issues exceeding a severity threshold (e.g., >10 occurrences in the last 15 minutes) using customizable search syntax.
2. Get Issue Details (Sentry) — Retrieves comprehensive issue data including full stacktraces, breadcrumbs, affected releases, and user impact metrics for deeper analysis.
3. Search Issues (Linear) — Performs full-text search across Linear to identify existing related issues and prevent duplicate ticket creation.
4. Claude Sonnet 4.6 Analysis (mcp-claude) — Leverages advanced LLM reasoning to generate root-cause hypotheses, identify affected code paths, and suggest targeted fix approaches.
5. List Teams (Linear) — Queries Linear workspace to identify the engineering team responsible for the affected service.
6. List Workflow States (Linear) — Retrieves available triage workflow states (e.g., Triage, Backlog) to properly categorize new issues.
7. Create Issue (Linear) — Creates a detailed Linear ticket with title, description, stacktrace, LLM-generated diagnosis, severity labels, and team assignment.
8. Create Attachment (Linear) — Attaches the original Sentry issue URL to the Linear ticket for cross-reference and source tracking.
9. Update Issue (Sentry) — Links the newly created Linear ticket back to the Sentry issue and marks it as In Progress.
10. Send Message (Slack) — Posts a formatted alert to the responsible team's Slack channel with error summary, Linear link, and Sentry link.

Key Features

• Intelligent Deduplication — Automatically detects and links duplicate errors to existing Linear issues instead of creating redundant tickets, reducing noise and improving team efficiency.
• AI-Powered Root Cause Analysis — Uses Claude Sonnet 4.6 to generate sophisticated diagnoses and fix suggestions directly from stacktraces, accelerating debugging and reducing time-to-resolution.
• Multi-System Integration — Seamlessly connects Sentry, Linear, and Slack into a unified error-handling workflow, eliminating context-switching and manual ticket creation.
• Automatic Team Routing — Identifies the owning engineering team and routes alerts to the correct Slack channel with full issue context, ensuring rapid response.
• Rich Issue Context — Captures complete error metadata (breadcrumbs, affected releases, user impact, LLM analysis, and original stacktrace) in a single Linear ticket for comprehensive incident understanding.

Use Cases

• Production Incident Response — Automatically surface critical production errors to engineering teams with AI-generated diagnostics, enabling faster root-cause identification and fixes in high-urgency scenarios.
• Error Triage Automation — Eliminate manual error log reviews and ticket creation by automatically categorizing, analyzing, and routing errors to the right teams with suggested solutions.
• Duplicate Error Consolidation — Prevent ticket spam by intelligently detecting recurring issues and linking them to existing Linear tickets, maintaining a single source of truth for each error class.
• Cross-Functional Issue Tracking — Maintain synchronized error records across Sentry and Linear with bidirectional updates, ensuring all stakeholders have current incident status and resolution progress.