Typed trace timeline
Agents SDK turns, tools, handoffs, guardrails, and MCP calls are first-class spans.
Flight recorder for OpenAI agents
Capture every Agents SDK run as a structured trace. Inspect the failed span, replay it against GPT-5.6, and turn the fix into a regression eval.
TRACE / REFUND AGENT
0.42s
0.31s
FAIL
{"amount": "12.OO"}
{"amount": 12}
FLIGHT PLAN / 01
One continuous path. No raw-log archaeology.
One tracing processor records turns, tools, handoffs, guardrails, and MCP calls as typed spans.
Read the exact prompt, arguments, output, timing, and failure—without scanning a wall of JSON.
Re-drive captured context while recorded tool outputs act as safe, side-effect-free stubs.
Generate structured assertions from the successful replay and run them as a regression eval.
INSTRUMENT PANEL / 02
Agents SDK turns, tools, handoffs, guardrails, and MCP calls are first-class spans.
Replay through one Responses API boundary and compare the original with the new output.
Turn fixed production behavior into structured, repeatable assertions.
Secret-like values are removed before trace payloads reach persistence.
See tool arguments, outputs, token use, and verdict changes in one review surface.
Run Phoenix and PostgreSQL on one box. Keep operational traces in your infrastructure.
RECORDER GUARANTEES / 03
READY FOR THE NEXT FAILURE?