# Sentry MCP for AI Agents AI Agent Connect > Sentry MCP for AI Agents. Monitor application errors and production health through natural conversation. Query issues, inspect raw events, audit release metadata, and check alert rules directly from your AI client to triage bugs faster. ## Overview - **Category:** loved-by-devs - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_UdHnFmprmwayCFV8FiFHkye9JqzNMUa8tL9yL63m/ai-agent-connect - **Tags:** error-tracking, performance-monitoring, application-health, debugging, alerting, issue-management ## Description Connect your Sentry account to your AI client and get real-time observability over your application errors without jumping between tabs. Instead of hunting through dashboards, you can ask your agent to find specific bugs, pull stacktraces, or summarize what went wrong in a recent deployment. This Connector lets you handle the tedious parts of on-call work, like checking breadcrumbs or identifying which release introduced a regression. You can audit your alert rules to cut down on noise or quickly see which team members are assigned to specific issues. By adding this to your workflow via the Vinkius Marketplace, you turn your AI into a dedicated engineer that knows your app's health inside out. It makes debugging a conversation rather than a search task. ## Tools ### get_event Get full details for a specific Sentry event. This includes the stacktrace, breadcrumbs, and HTTP context. ### get_issue Get full details for a Sentry issue using its numeric ID. Use this to see the history and status of a bug. ### get_project Get details for a specific Sentry project. This shows the configuration and ownership of an application. ### get_release Get details for a specific Sentry release. Use this to correlate bugs with a specific version string. ### list_issues List issues in a Sentry organization or project. You can filter by status, priority, or text search. ### list_projects List projects in a Sentry organization. Use this to see the full footprint of your monitored apps. ### list_releases List releases for a Sentry organization or project. This tracks which versions have been deployed. ### search_issues Search Sentry issues by text using Sentry query syntax. Find specific bugs across your entire organization. ### list_tags List tags for a Sentry organization or project. These are essential for filtering and grouping your errors. ### list_teams List teams in a Sentry organization. Use this to see member access and team ownership. ### update_issue Update a Sentry issue status or assign it to a team member. Use this to manage your bug backlog directly. ### list_events List recent events for a Sentry project. This is great for auditing what errors have been firing lately. ### list_alert_rules List all alert rules in a Sentry organization to see how notifications are triggered. This helps you manage Slack and PagerDuty pings. ### get_auth_info Verify your Sentry token is working correctly. Use this to check permissions and connection status. ### list_organizations List all Sentry organizations. This helps you get the slugs needed for other API calls. ## Prompt Examples **Prompt:** ``` Show me the most recent errors for the production api. ``` **Response:** ``` I've pulled the recent events for your production api. Here are the latest issues: | Error Message | Frequency | Environment | | :--- | :--- | :--- | | `NullPointerException` in AuthModule | 42 | production | | `TimeoutException` in DatabasePool | 12 | production | Would you like me to pull the full stacktrace for that `NullPointerException`? ``` **Prompt:** ``` Which release introduced the 'Payment Gateway' bug? ``` **Response:** ``` I've analyzed the issues related to 'Payment Gateway'. * **Bug identified:** `GatewayTimeoutError` in the checkout flow. * **First seen in release:** `v2.4.1` * **Deployment date:** October 12th. I can also pull the deployment metadata for `v2.4.1` if you need to see who deployed it. ``` **Prompt:** ``` Are there any noisy alert rules for the mobile team? ``` **Response:** ``` I've audited the alert rules for the mobile team. I found 3 rules that might be too noisy: 1. **Slack Alert:** Fires on every `Warning` level log (currently firing 50+ times an hour). 2. **Email Alert:** Sends an email for every single `AuthFailure` (high frequency). 3. **PagerDuty:** Triggers on `SlowResponse` (threshold is set too low). Would you like me to show the full configuration for any of these? ``` ## Capabilities ### Triage unresolved issues Query your backlog to find and prioritize bugs based on status, priority, or text. ### Inspect raw error events Get full stacktraces and breadcrumbs to see exactly what happened before a crash. ### Audit alert rules Review your Slack, email, and PagerDuty triggers to manage notification noise. ### Track application releases Correlate specific errors with deployment versions to find regressions quickly. ### List organization projects Get a high-level view of all projects and teams across your Sentry account. ### Analyze event tags View tags for environments and transactions to filter errors more effectively. ## Use Cases ### Quick Triage for PagerDuty Alerts When an alert fires, ask your agent to use get_issue to see the error and get_event to pull the breadcrumbs and stacktrace. ### Audit Alert Noise An engineering manager uses list_alert_rules to find out why the team is getting too many Slack pings and identifies noisy rules. ### Identify Regression Versions A developer uses search_issues to find a new bug and then calls get_release to see which deployment first introduced it. ### Onboarding New Engineers Use list_projects and list_teams to give a new hire an instant overview of the app's error landscape and ownership. ## Benefits - Stop dashboard fatigue by using list_issues to see what's broken without opening a browser tab. - Find root causes faster with get_event which brings stacktraces and breadcrumbs straight to your chat. - Audit your notification pipeline with list_alert_rules to reduce alert noise for your team. - Verify deployments with list_releases to see exactly which version introduced a regression. - Manage your backlog using update_issue to resolve or assign bugs while you're still in the chat. - Get a high-level overview of your error landscape using list_projects to see all monitored apps at once. ## How It Works The bottom line is you get a chat-based interface for your entire Sentry error tracking pipeline. 1. Subscribe to this Connector on the Vinkius Marketplace. 2. Provide your Sentry Internal Integration Token in your client settings. 3. Ask your AI client to check for production errors or summarize recent release issues. ## Frequently Asked Questions **Can the Sentry MCP help me triage production errors faster?** Yes, it lets your AI client pull real-time error data so you can see stacktraces and breadcrumbs instantly without leaving your chat interface. **How does Sentry MCP help with on-call rotations?** It acts like a dedicated engineer that can check your issue backlog and tell you what needs attention right now, helping you prioritize your work. **Can I use Sentry MCP to see which release broke my app?** Yes, you can use it to correlate specific bugs with deployment versions to identify exactly which release introduced a regression. **Does Sentry MCP support searching for specific bugs?** You can use natural language to search for issues across your projects. The AI will handle the query syntax to find the bugs you need. **Can I manage my Sentry issues through an AI agent?** You can use the Connector to change statuses or assign bugs to team members directly from your chat, making it easier to manage your backlog. **Is the Sentry MCP good for auditing alert rules?** It is excellent for that. You can use it to see exactly how and where your team is being notified to help reduce alert fatigue. **How do I create a Sentry Internal Integration Token?** Go to your [**Sentry Organization Settings**](https://sentry.io/settings/), select **API Keys** > **Create New Integration**. Name it (e.g. 'Vinkius MCP'), select the scopes you need (recommended: `org:read`, `project:read`, `event:read`, `team:read`, `release:read`), and click **Save**. Copy the token immediately — it starts with `sntrys_` and won't be shown again. **Can I resolve or mute an issue directly from the agent?** Yes! Use the `update_issue` action with the issue's numeric ID and set `status` to `resolved` (marks it fixed), `muted` (hides it from the inbox) or `deleted` (removes it permanently). You can also reassign issues using the `assigned_to` parameter. **What Sentry query syntax does the search support?** Sentry uses its own query syntax. Common filters include: `is:unresolved` (open issues), `is:resolved` (fixed issues), `priority:[1-50]` (priority levels), `first_release:1.2.3` (issues introduced in a specific release), `user.email:example@co.com` (errors affecting a specific user). You can combine filters: `is:unresolved priority:50`.