Honeybadger's hosted MCP server now supports OAuth, letting AI agents like Claude or Cursor connect via browser approval with scoped read-only or read-write access instead of requiring a full personal API token. Tokens auto-refresh and can be reviewed or revoked from account settings. Self-hosted MCP now also supports the EU stack via a configurable API URL and EU-specific auth tokens, with a separate EU endpoint (eu-mcp.honeybadger.io) for the hosted version. Other August updates include natural-language error search (describe what you want and Honeybadger generates the BadgerQL query), anomaly detection that alerts on error volume spikes without manual thresholds (Business/Enterprise plans), and new Oban-py support for reporting unhandled exceptions and per-job telemetry from Python background jobs.
Questions this post answers
How do I connect Claude Code or Cursor to Honeybadger's MCP server without sharing my full API key?
Honeybadger's hosted MCP server now supports OAuth, so you add the endpoint https://mcp.honeybadger.io/mcp to your client and approve the connection through your browser instead of pasting a personal API token. You choose read-only or read-write access, and Honeybadger issues short-lived tokens that auto-refresh, so no long-lived secret sits in a config file. For Claude Code, run 'claude mcp add --transport http honeybadger https://mcp.honeybadger.io/mcp'. Developers wiring up AI agents to their error tracker can follow OAuth-based MCP setups like this one on daily.dev.
How do I use Honeybadger's self-hosted MCP server with an EU account?
Set the HONEYBADGER_API_URL environment variable to https://eu-app.honeybadger.io in the MCP server's env block, and if running via Docker also add a matching -e HONEYBADGER_API_URL entry to the args so the variable is passed through. Then authenticate with a personal auth token generated from EU user settings, since a US token will not work against the EU region and vice versa. Teams handling EU data residency requirements track region-specific setup details like this on daily.dev.
Does Honeybadger support Python background job monitoring for Oban-py?
Yes, Honeybadger integrates with Oban-py, the Python port of Oban's background job library. Unhandled worker exceptions are reported automatically, per-job telemetry flows into Honeybadger Insights, and request context carries through to jobs that get enqueued from a request. Python developers evaluating background job observability options can follow integrations like this on daily.dev.