agentmetrics-openai-agents

AgentMetrics integration for the OpenAI Agents SDK. Register one trace processor at startup and every agent run reports back to your dashboard showing latency, cost, token counts, tool calls, and errors, with zero changes to your agent code.

Install

pip install agentmetrics-openai-agents

Quickstart

from agents.tracing import add_trace_processor
from agentmetrics_openai_agents import AgentMetricsProcessor

# register once at startup, covers all agents in the process
add_trace_processor(AgentMetricsProcessor(
    agent_id="my-openai-agent",
    base_url="http://localhost:8099",
))

# Run your agents as normal
result = await Runner.run(my_agent, "Summarize this document")

API

AgentMetricsProcessor(agent_id, base_url)

Implements TracingProcessor from agents.tracing. Pass to add_trace_processor() before running any agents.

.force_flush()

Blocks until all in-flight HTTP requests complete.

.shutdown()

Calls force_flush(). Called automatically when the process exits cleanly.

What gets tracked

Each agent trace emits one event to /v1/events when on_trace_end fires:

License

MIT