Sentry

What it does

This skill provides read-only access to Sentry, a platform for application error tracking and performance monitoring. It allows an AI agent to inspect production errors, retrieve issue details, view events associated with those issues, and examine individual event data – all while automatically redacting personally identifiable information (PII). The skill leverages the Sentry CLI and handles authentication, pagination, and retries behind the scenes for simplified querying.

When to use it

• Troubleshooting Production Issues: Quickly identify and investigate unresolved errors in a production environment.
• Analyzing Error Trends: Examine recent issues and their details to understand patterns and potential root causes.
• Reviewing Event Data: Inspect individual error events, with PII automatically removed for privacy compliance.

Key capabilities

• Issue Listing: Retrieves a list of unresolved issues, ordered by recency.
• Issue Details Retrieval: Fetches detailed information about specific Sentry issues.
• Event Inspection: Allows viewing the events associated with an issue.
• PII Redaction: Automatically removes personally identifiable information from event data.
• Automatic Pagination & Retries: Handles large datasets and transient errors automatically.

Example prompts

• "List the 20 most recent unresolved issues in production."
• "What are the details of issue short ID '12345'?"
• "Show me the events associated with issue '67890'."

Tips & gotchas

• Authentication: Requires a read-only SENTRY_AUTH_TOKEN with project:read, event:read, and org:read scopes. Do not paste your token directly into chat; set it as an environment variable instead.
• Auto-Detection: The skill attempts to automatically detect the Sentry organization and project. If this fails, you'll need to explicitly specify them in your prompts.
• JSON Output: Use --json when processing output programmatically for structured data.