← Back to Cookbook

Debug and resolve issues with Sentry MCP and Warp

Connect the Sentry MCP server to Warp so your AI terminal agent can pull issues, read traces and logs, call Seer for a root-cause analysis, and apply the fix.

Features
Category Debugging
Time
10–15 minutes
Difficulty
Beginner
Steps
4 steps

Before you start

Tools
  • Warp installed and running
  • Node.js 18+ — only needed for the command-based mcp-remote fallback
Accounts & access
Knowledge
  • Basic familiarity with Warp and its Agent Mode chat input

1
Install and authenticate the Sentry MCP server

In Warp, MCP servers are added through the settings UI. Open Settings > Agents > MCP Servers, Select the Sentry from the list of recommend servers by pressing the '+' icon. This will automatically trigger to OAuth flow for you to sign into your Sentry account and enable access to Sentry resources.

Sentry MCP server documentation

2
Ask Warp to summarize your open issues

Switch Warp into Agent Mode by entering the command '/agent' and ask it to pull issues from your Sentry project. Warp will use the Sentry MCP tools to list current issues, grouped by frequency and recency. Try a prompt like the one in the video and adjust the project slug and timeframe to match your setup. You can find your project slug in Sentry under Settings > Projects.

3
Run a Seer root-cause analysis

Once you spot an issue worth investigating, ask Warp to investigate the issue further with Seer. Seer is Sentry's AI debugging agent. It analyzes the error details and stack trace, distributed traces, structured logs, and profiles, cross-references your linked codebases, and produces a step-by-step root-cause analysis with the most likely culprit and a suggested fix.

  • Root cause — the sequence of events that led to the issue and the most likely culprit, annotated with links to the exact lines of code
  • Supporting evidence — the traces, logs, and telemetry Seer used to reach its conclusion
  • Suggested fix — a concrete code change you can review, refine, and apply
Seer root cause analysis documentation

4
Apply the fix

With Seer's root cause and proposed fix in hand, ask Warp to implement the changes in your codebase. Because Warp has access to both your codebase and the Sentry issue context with Seer's analysis, it can write a targeted, accurate patch rather than a generic suggestion.

Review the diff before accepting. Once you're happy, commit and deploy as usual. Check Issues in Sentry to confirm the error rate drops after the fix ships.

That's it.

Seer found the cause. Warp shipped the fix.

With Sentry's MCP server connected to Warp, your AI terminal agent can search errors, traces, and logs, trigger a Seer root-cause analysis, and apply the fix in a fraction of the time it would take manually.

  • Install and authenticate the Sentry MCP server in Warp
  • Ask Warp to list and summarize open issues from your Sentry project
  • Analyze an issue with Seer using Sentry stack traces, structured logs, and related spans
  • Apply the fix directly using a Warp agent

Pro tips

  • 💡 Give Warp a specific issue ID (e.g. MYAPP-123) rather than a description. It removes ambiguity and lets Warp fetch the exact event data — and run Seer against it — in one tool call.
  • 💡 After Seer returns a root cause, ask Warp: Are there any other Sentry issues in this project that likely share this root cause? This surfaces systemic problems, not just the one you're looking at.
  • 💡 Combine Seer's analysis with your git history: Show me recent commits to the file Seer flagged. Warp can correlate the root cause with recent changes to narrow down when the bug was introduced.

Common pitfalls

  • ⚠️ Complete the OAuth flow before asking Warp to fetch Sentry data. If you skip it or dismiss the browser prompt, Warp won't have access to your organization and MCP tool calls will fail.
  • ⚠️ Connect your GitHub repository to Seer before relying on its root-cause analysis. Without a linked codebase, Seer can still reason over telemetry, but its culprit line references and suggested fix are far more accurate with repo access — set it up under Seer settings.
  • ⚠️ Don't apply Seer's suggested fix blindly. It's a strong starting point, but validate the diagnosis against your own knowledge of the codebase and review the diff before committing.

Frequently asked questions

No. The Sentry MCP server works with any Sentry plan, including the free tier. Access to specific data (e.g. longer retention windows, Seer root-cause analysis) depends on your plan, but the core issue-fetching and log-reading tools are available to all users.

Seer works best with a GitHub repository connected in your Seer settings. With a linked codebase it can point to the exact culprit lines and generate a concrete fix; without one, it still reasons over your issue details, traces, logs, and profiles to explain the likely root cause.

Yes, but the setup differs. If you're running a self-hosted Sentry installation, use the command-based mcp-remote (or STDIO) transport in Warp instead of the hosted URL. See the Sentry MCP documentation for full self-hosted setup instructions.

Yes. The Sentry MCP server supports 12+ clients, including Claude Code, Cursor, Windsurf, and any tool that implements the Model Context Protocol. The setup steps differ slightly per client. Refer to the Sentry MCP docs for client-specific instructions.

When Warp fetches data via the Sentry MCP server, that data becomes part of your Warp conversation context and is processed by the AI model Warp is configured to use. Review Warp's and Sentry's privacy and data-usage documentation if you have compliance requirements before connecting production data.

Fix it, don't observe it.

Get started with the only application monitoring platform that empowers developers to fix application problems without compromising on velocity.