Sentry is a Monitoring MCP server that lets Claude Code, Cursor, Windsurf and any MCP-compatible AI agent error tracking and performance monitoring. Install in 1 minute with mcpizy install sentry.
Connect your Sentry account once — MCPizy stores the credentials encrypted and uses them whenever you run a recipe in managed mode.
mcpizy install sentrynpx -y @sentry/mcp-serverlist_issuesList issues in a project
Inputs
organization_slugstringrequiredproject_slugstringrequiredget_issueGet full details about an issue
Inputs
issue_idstringrequiredlist_projectsList projects in the organisation
search_errorsSearch errors by query
Inputs
querystringrequiredget_eventGet a specific event (stacktrace etc.)
Inputs
event_idstringrequiredWorks identically across clients. Only the config file path differs.
~/.claude.json{
"mcpServers": {
"sentry": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.sentry.dev/mcp"
]
}
}
}.cursor/mcp.json{
"mcpServers": {
"sentry": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.sentry.dev/mcp"
]
}
}
}~/.codeium/windsurf/mcp_config.json{
"mcpServers": {
"sentry": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.sentry.dev/mcp"
]
}
}
}Authenticate via OAuth on first connect
Paste any of these prompts into Claude Code, Cursor or another MCP-compatible client.
“Show me unresolved production errors from the last 24h”
Uses: list_issues
“What's the stacktrace for issue SENTRY-42?”
Uses: get_issue
“Which errors spiked in the last release?”
Uses: search_errors
Sentry MCP exposes the Sentry API as agent tools: `list_projects`, `list_issues`, `get_issue_details`, `list_events_for_issue`, `resolve_issue`, `assign_issue`. The point is to let an AI walk your error inbox the way an engineer would — open the top issue, read the stack trace, fetch the breadcrumbs and request payload, decide whether it's a real bug, file a Linear ticket, and resolve or ignore. The official Sentry MCP is OAuth-based; community forks accept a personal auth token.
We use Sentry MCP daily for on-call rotations. The agent runs every morning: "open Sentry, find the issues that spiked in the last 24h vs the previous 7-day baseline, group by file, and post a Slack summary." That single workflow replaces a 20-min manual triage. The second use is fix-loop pairing — when Claude Code is debugging a regression, "look at Sentry for crashes in this function in the last 7 days" gives the agent real production stack traces with line numbers. Token cost is heavier than average: a single issue with 5 events and full breadcrumbs runs ~4-6k tokens because Sentry events carry request payloads, user context, and source maps.
Compared to Datadog or Honeycomb MCPs (when they ship), Sentry remains the cleanest for application-level error tracking with source-mapped stack traces. For broader observability (logs, traces, metrics), Sentry is shallower than Datadog. The honest trade-off: the MCP is excellent for issue triage and worse for ad-hoc queries — Sentry's search syntax (`event.type:error environment:production level:error`) is not always exposed cleanly through the MCP, so complex filters still need the web UI.
Project scoping is per-token. A token created in one Sentry team can't see another team's projects, and the error message is generic. Generate the token at the organisation level if you want the agent to triage across teams.
Event payloads can be large (50-200 kB per event with source-map context). Asking the agent to "fetch all 50 events for this issue" will blow past the model's context window. Limit to the top 3-5 events per issue.
`resolve_issue` does not auto-create a Linear/Jira ticket — it just flags the issue as resolved in Sentry. Chain the resolution to a separate Linear/GitHub MCP call to keep the trail.
Rate limits: 40 req/min for most read endpoints. An agent that fetches details for 50 issues sequentially will saturate the limit. Batch with `list_issues` first and only fetch full detail for the top 5-10.
Honest pros/cons against the closest monitoring MCP servers.
| Server | Strengths | Trade-offs |
|---|---|---|
| Datadog MCP (community) | Broader observability — logs, metrics, traces in one place | Less code-aware than Sentry, no source-mapped stacks |
| Rollbar MCP (community) | Cheaper for small teams, similar error grouping | Smaller ecosystem, fewer maintained MCP forks |
| Honeycomb MCP | Best for high-cardinality trace queries | Tracing-first, weak as an error inbox |
Access OpenTelemetry traces and metrics via Pydantic Logfire. Debug production issues fast.
Crash reporting and real user monitoring data. Track errors, performance, and user sessions.
Real-time production context — logs, metrics, traces. SRE intelligence for incident response.
Query metrics and alerts
If Sentry doesn't fit your stack, these Monitoring MCP servers solve similar problems.
The Sentry MCP server is an Monitoring Model Context Protocol server that lets Claude Code, Cursor, Windsurf, VS Code with Copilot, and other MCP-compatible AI agents error tracking and performance monitoring. It exposes Sentry's capabilities as tools the AI can call directly from your editor or CLI.
The fastest way is the MCPizy CLI: run `mcpizy install sentry` and MCPizy will add the server to your `.claude.json` automatically. You can also install it manually by adding an entry under `mcpServers` in `.claude.json` with the command `npx -y @sentry/mcp-server` and restarting Claude Code.
Yes. The Sentry MCP server is free and open source (see the GitHub repository linked on this page). You may still need a Sentry account or API key to connect the server to the underlying service, but the MCP layer itself has no MCPizy subscription cost.
Yes. Any MCP-compatible client works — including Claude Code, Claude Desktop, Cursor (via `.cursor/mcp.json`), Windsurf, VS Code with Copilot Chat, and custom agents built on the MCP SDK. The same install command targets all of them; only the config file path differs.
Once installed, your AI agent can error tracking and performance monitoring directly inside your conversation. Typical use cases include asking Claude Code or Cursor to run Sentry operations, inspect results, chain Sentry with other MCP servers (see our Workflow Recipes), and automate repetitive monitoring tasks without leaving your editor.