Find what went wrong in your LLM app with Phoenix.
Guides developers in adding Phoenix tracing, evaluations, and monitoring to LLM applications.
When to use it
Use when debugging LLM applications, evaluating datasets, monitoring production systems, or comparing prompts and models.
Give it an LLM application to observe; it adds Phoenix instrumentation and evaluation code to the project.
What you provide
The server-launch and integration examples run in Python; no Python version is specified.
Enables the optional authenticated server setup as its JWT signing secret, alongside PHOENIX_ADMIN_SECRET.
This skill
Phoenix
Starts a Phoenix server
Records evaluations and deletes old traces (irreversible)
Slack
Sends production error alerts (irreversible)