agent-observability-trace-rca
datadog-labs515 installs
Find why your production LLM app is failing.
What it does
Diagnoses failures in production LLM applications and returns an evidence-backed report with specific fixes.
When to use it
Use when investigating application failures, failing evaluations, runtime errors, or production failure patterns.
How to use it
Give it an application name; it investigates production traces and returns a root-cause analysis report in chat.
What you provide
- the application name
Uses
Good to know
- acts without asking
Latest review
What you need · 2
Uses the Datadog MCP server with the llmobs toolset on the preferred backend path; pup is an alternative.
Requires authenticated access to Datadog; the CLI path stops and asks the user to log in on authentication errors.
Access · 1
This skill
Datadog
Write
Sends analysis intent telemetry