EVALS V1.3.0: OPENAI AGENTS SUPPORT, STRICT JSON WRITES, AND UTC TIMESTAMPS
strands-agents/evals v1.3.0 adds OpenAI Agents support and tightens JSON and time handling for more reliable eval pipelines. The [release notes](https://github...
strands-agents/evals v1.3.0 adds OpenAI Agents support and tightens JSON and time handling for more reliable eval pipelines.
The release notes add OpenAI Agents support to the GenAI session mapper, so you can evaluate agent runs without bespoke adapters.
It also rejects non‑strict JSON on write, skips non‑serializable tools during serialization, and normalizes timestamps to timezone‑aware UTC—plus safer model serialization defaults.
As voice and tool‑using agents push latency budgets, correctness in eval logging matters; see The New Stack’s overview for why timing and structure are critical.
Agent evals get first-class support for OpenAI Agents sessions, reducing glue code.
Strict JSON and UTC timestamps improve reproducibility, audits, and downstream analytics.
-
terminal
Run the GenAI session mapper against OpenAI Agents runs; verify tool calls, messages, and metrics map as expected.
-
terminal
Re-export experiments and logs; confirm strict JSON rejects invalid values and all timestamps are timezone-aware UTC.
Legacy codebase integration strategies...
- 01.
Expect some writes to fail if prior workflows emitted non‑strict JSON; add validation or cleanup before persisting.
- 02.
Rebuild dashboards that assumed local time or naive datetimes; standardize on UTC parsing.
Fresh architecture paradigms...
- 01.
Adopt v1.3.0 as the baseline for agent evals to get OpenAI Agents support out of the box.
- 02.
Design schemas with strict JSON and UTC from day one to avoid ingestion drift and replay mismatches.
Get daily OPENAI-AGENTS-SDK + SDLC updates.
- Practical tactics you can ship tomorrow
- Tooling, workflows, and architecture notes
- One short email each weekday