Features
Automatic trajectory capture for every OpenClaw agent run
Interactive trace tree with LLM calls, tool use, and sub-agent payloads
Call graph visualization of agent-tool-model relationships
Timeline Gantt chart to spot bottlenecks and parallelism
Trajectory dashboard with metrics, daily trends, and token usage
Tracy doctor agent – ask natural-language questions about failures and costs
Per-step cost attribution with model-specific pricing
Error flagging and efficiency scoring
Live data queries without stale dashboards
Conversation memory with Tracy for follow-up questions
OpenClaw plugin integration installable via CLI
Credits-based usage billing with no seat commitments
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
