Error Tracking

Error Tracking wires Arthur's backend into the error-monitoring provider your team already uses, instead of leaving failures to only show up in raw server logs.

What you can do

• Forward HTTP request errors, MCP tool/chain/resource/prompt execution failures, and uncaught process-level exceptions.
• See failures with useful context — route, error source, status code — so they're actionable from your provider's dashboard.
• Trust that request bodies, cookies, authorization headers, API keys, connection strings, and secret values are never forwarded. The goal is visibility into failures, not a second place your credentials could leak from.

How to use it

1. Go to Error Tracking and add the provider you use.
2. Configure the connection details for that provider.
3. Trigger a test failure or wait for a real one, and confirm it shows up on your provider's dashboard.

Good to know

• Error Tracking is about unexpected runtime failures. Deliberate changes made by people are tracked separately in Audit Logs.