Arize-aiOfficial·36 Agent Skills Included

phoenix

Trace, evaluate, and debug LLM applications in one platform

Captures detailed traces of LLM calls, tool executions, and retrievals using OpenTelemetry instrumentation. Scores output quality with code-based and LLM-as-a-judge evaluators, and runs dataset experiments to compare prompts and models. Eliminates blind debugging of AI agents by surfacing failures, costs, and latency in a searchable UI and CLI.

11,277|1,085

↗

npx skills add Arize-ai/phoenix --all -g -y

All Skills in This Repository (36)

Pure Emerald Level Indicators

📦 In Repo

Arize-ai

phoenix-cli-development

Guides design and implementation of noun-verb commands for the Phoenix CLI.

phoenix-client-development

Guides development of the phoenix-client TypeScript SDK covering experiments, prompts, tracing, and vitest testing.

phoenix-otel-development

Guides OpenTelemetry registration and global tracer provider lifecycle management in TypeScript.

phoenix-server

Guides backend development of Phoenix's FastAPI, Strawberry GraphQL, and SQLAlchemy server codebase.

agent-browser

Automates browser interactions via a CLI using Chrome DevTools Protocol and accessibility-tree snapshots.

phoenix-skills-audit

Audit recent Phoenix client, CLI, and API changes and patch stale agent skill documentation.

pxi-eval-dataset

Generate synthetic YAML evaluation datasets for the PXI eval harness.

mintlify

Build and maintain Mintlify documentation sites with MDX pages and docs.json configuration.

gh-stack

Manages stacked GitHub pull requests and splits multi-part work into reviewable branch layers.

phoenix-sqlean

Maintains the vendored sqlean.py fork by bumping pinned SQLite, sqlean, and xxHash versions.

phoenix-integration-snippets

Generates onboarding code snippets for Phoenix tracing integrations and wires them into the onboarding UI.

phoenix-typescript

Enforces TypeScript naming, typing, and import conventions across the Phoenix monorepo.

Official

Intermediate

Frequently Asked Questions

FAQPage Schema

How to install Phoenix?▼

Run `npx skills add Arize-ai/phoenix --all -g -y` in your terminal to install all skills in this suite globally.

What is Arize Phoenix used for?▼

Phoenix is an open-source AI observability platform for tracing, evaluating, and debugging LLM applications. It helps you see exactly what your AI agents and prompts are doing in production.

How do I trace my LLM app with Phoenix?▼

Install the Phoenix OTel package for Python or TypeScript, call register() with your project name, and your framework's LLM calls are captured automatically. Traces appear in the Phoenix UI at localhost:6006.

Does Phoenix work with Claude Code and Cursor?▼

Yes. Phoenix ships a remote MCP server and CLI that connect directly to Claude Code, Cursor, Codex, and other MCP-compatible agents for querying traces and experiments.

Can Phoenix evaluate RAG and agent quality?▼

Yes. It includes pre-built evaluators for faithfulness, relevance, and tool selection, plus custom code or LLM-as-a-judge evaluators you can run over traces and datasets.