Features
Real-time MCP request tracing (tool calls, prompts, resources)
Full JSON-RPC payload capture with arguments, results, errors
Per-server/per-client/per-tool performance metrics (P50/P95/P99)
Session tracking and duration breakdowns
Error tracking with stack traces and context
Alert rules (Pro: up to 10, Business: up to 100) with Slack, webhook, email
Public dashboards (Pro+) with read-only share links
Multi-region ingestion (US and EU)
Data retention from 30 days (Free) to 12 months (Business)
SAML/OIDC single sign-on (Business)
Open-source SDK and CLI (Apache 2.0)
No-code install via CLI (wraps stdio/HTTP proxy)
W3C trace context propagation for APM correlation
Header redaction for credentials
Trace visibility for agent steps (prompts, retrievals, tool calls, outputs)
LLM-as-judge evaluation for relevance, toxicity, quality scoring
Dataset creation from traces for reproducible testing
Experiment management and regression benchmarking
Built-in Prompt IDE for iterative prompt optimization
Self-hosted deployment on local, Docker, Kubernetes
Phoenix Cloud managed hosting option
Vendor-agnostic support for any model/framework
Native OpenTelemetry integration
OpenInference specification for LLM telemetry
Human annotation and automated labeling
Ghost trajectories to simulate alternative agent paths
Eval-as-you-test for early quality feedback
One-click integration with LlamaIndex
