Skip to main content
Phoenix provides tools for AI coding assistants to debug and analyze your LLM applications directly from the terminal.

​
Installation

​
Phoenix CLI

Install the CLI globally: 
npm install -g @arizeai/phoenix-cli

​
Phoenix Skills

Add Phoenix skills to your AI coding agent using skills add: 
npx skills add Arize-ai/phoenix
This installs the skill to your project’s skills directory (.cursor/skills/, .claude/skills/, or .github/skills/).

​
Available Skills

SkillDescriptionRecommended
phoenix-cliDebug and analyze LLM applications using Phoenix CLI tools. Teaches agents to fetch traces, analyze performance, and review experiments.Yes
phoenix-evalsBuild and run evaluators for AI/LLM applications. Teaches agents to create code and LLM evaluators, run experiments, and validate against human feedback.Optional
phoenix-tracingOpenInference semantic conventions and instrumentation. Teaches agents proper span types, attributes, and production deployment patterns.Optional

​
skills add Options

OptionDescription
-g, --globalInstall to user directory instead of project
-a, --agent <agents...>Target specific agents (e.g., claude-code, cursor)
-s, --skill <skills...>Install specific skills by name
-l, --listList available skills without installing
-y, --yesSkip all confirmation prompts
Examples: 
# Install the CLI skill globally for all projects
npx skills add Arize-ai/phoenix --skill phoenix-cli -g

# Install both skills for comprehensive Phoenix support
npx skills add Arize-ai/phoenix --skill phoenix-cli --skill phoenix-tracing

# Install to specific agents
npx skills add Arize-ai/phoenix --skill phoenix-cli -a claude-code -a cursor

# Non-interactive installation (CI/CD friendly)
npx skills add Arize-ai/phoenix --skill phoenix-cli -g -y
Supported agents: Claude Code, Cursor, Windsurf, Codex, GitHub Copilot, Cline, OpenCode, Gemini CLI, and 20+ more.

​
Configuration

Set environment variables to connect to your Phoenix instance: 
export PHOENIX_HOST=http://localhost:6006    # Your Phoenix endpoint
export PHOENIX_PROJECT=my-project            # Project name
export PHOENIX_API_KEY=your-api-key          # API key (if auth enabled)

​
Using the Skills

​
phoenix-cli

The CLI skill teaches your AI assistant how to:
  • Debug failed LLM applications using trace data
  • Analyze performance issues and token usage
  • Review experiment results and datasets
Example prompt: 
Debug my agent - it's failing on tool calls. Fetch recent traces and identify the issue.

​
phoenix-tracing (Optional)

The tracing skill teaches your AI assistant how to:
  • Instrument LLM applications with OpenInference tracing
  • Create custom spans with proper attributes (LLM, RETRIEVER, TOOL, AGENT, etc.)
  • Follow semantic conventions for consistent observability
  • Configure production deployment patterns (batching, PII masking)
Example prompt: 
Add Phoenix tracing to my RAG pipeline. Create spans for the retriever and LLM calls.

CLI Reference

Full CLI command reference

Retrieve Traces via CLI

Detailed guide for fetching traces

MCP Servers

Connect AI assistants to Phoenix via MCP