sentry-automation
Automate Sentry Tasks with Claude
This skill enables Claude to manage Sentry error tracking operations including issue investigation, alert configuration, release tracking, and cron monitoring through Composio Rube MCP integration.
スキルZIPをダウンロード
Claudeでアップロード
設定 → 機能 → スキル → スキルをアップロードへ移動
オンにして利用開始
テストする
「sentry-automation」を使用しています。 List unresolved errors in my backend-api project
期待される結果:
Found 23 unresolved issues in backend-api:
🔴 Critical (3):
- Connection timeout in Redis client (seen 234 times)
- Memory leak in worker process (seen 89 times)
- Uncaught promise rejection (seen 67 times)
🟡 Warning (8):
- Deprecated API endpoint called (seen 45 times)
- Slow database query > 2s (seen 32 times)
...
Would you like me to investigate any specific issue?
「sentry-automation」を使用しています。 Create an alert for high error rate
期待される結果:
Alert rule created successfully:
Name: High Error Rate - backend-api
Condition: Error rate > 5% in 10 minutes
Actions: Email to on-call team, Slack #alerts
Project: backend-api
Status: Active
The alert will start monitoring immediately.
セキュリティ監査
安全This skill provides legitimate automation for Sentry error tracking through Composio Rube MCP. The static findings flagged as external_commands are false positives - they detect MCP tool names in documentation (e.g., SENTRY_LIST_AN_ORGANIZATIONS_ISSUES), not shell execution. The network reference is to a documented MCP server endpoint (rube.app/mcp) that users explicitly configure. No actual security risks identified.
低リスクの問題 (2)
リスク要因
🌐 ネットワークアクセス (1)
品質スコア
作れるもの
Production Error Investigation
Developer uses Claude to quickly find and investigate critical errors in production by querying Sentry issues with specific filters.
Alert Rule Configuration
DevOps engineer creates and updates alert rules for new projects or modified thresholds using natural language commands.
Release Monitoring
SRE monitors new releases and their associated errors by tracking deployment events and issue correlations.
これらのプロンプトを試す
Use the sentry-automation skill to list the 10 most recent unresolved errors in my production project. Show the issue title, frequency, and last seen timestamp.
Create a new alert rule for my backend-api project that triggers when error rate exceeds 5% over 10 minutes. Notify the on-call team.
Create a new release version 2.1.0 for my organization, mark it as deployed to production environment, and upload source maps from the dist folder.
Search all projects in my organization for issues tagged with 'database' that have occurred more than 50 times in the last 24 hours. Group them by project and suggest priority.
ベストプラクティス
- Always call RUBE_SEARCH_TOOLS first to verify current tool schemas before workflow execution
- Verify Rube MCP connection status shows ACTIVE before running any Sentry operations
- Use specific query filters (is:unresolved, assigned:me, times-seen:>10) to reduce API calls and improve response time
回避
- Do not hardcode organization or project names - always use lookup tools to get current slugs
- Do not skip connection status verification - operations will fail if MCP connection is not active
- Do not ignore rate limit responses - implement backoff to avoid throttling