Sentry 201 · Workshop toolkit

Give your coding agent your production context, let it find the root cause and a fix, and let the pull request open itself.

Companion repo to Sentry 201, Session 3 — Better agent performance with Sentry context & AI tools. Everything we ran live is here, parameterized so you can drop it into your own Sentry project.

The path

Step	What the agent can do	Sentry pieces
01 · Setup	See production — issues, traces, logs, suspect commits	Sentry CLI (`sentry init`) · MCP
02 · Plugin	Work with Sentry the right way	`sentry-for-ai` skills
03 · Cursor automations	Fix on a schedule or when feedback lands, open PRs	Sentry MCP · Cursor Automations
04 · Claude routine	The same scheduled sweep, the Claude Code way	Sentry MCP · Claude scheduled tasks

More recipes: cookbook-links.md

Before you start

A Sentry project with an SDK installed and data flowing in
Repo connected for code mapping, and Seer enabled for the project (Settings → Seer)
One of: Claude Code or Cursor
gh CLI authenticated (gh auth login) if you want agents to open PRs

Human in the loop

Every workflow here stops at a pull request. Nothing merges on its own — the agent does the work, you do the review.

Config for your environment

Replace {MCP_ENDPOINT_URL} with this MCP’s endpoint URL (from its repo or docs above). No API key — you connect directly.

Tool

Config file: ~/.cursor/mcp.json

{
  "mcpServers": {
    "mcp-server": {
      "url": "{MCP_ENDPOINT_URL}"
    }
  }
}

Paste into mcpServers in the config file. Restart Cursor after saving.

sentry-201-workshop

Links

README

Sentry 201 · Workshop toolkit

The path

Before you start

Human in the loop

Collected info

Config for your environment