Summary
agent-learning 0.9.0 adds policy audit and replay with consumed-input lineage. The release records which inputs contributed to a policy execution so the run can be reproduced and examined later.
Why it matters
Agent developers can debug policy changes and investigate regressions with stronger provenance than ordinary logs or aggregate evaluation scores provide.
Related coverage
- OpenAI Node SDK 7.20.0 adds vault credentials and safety lifecycle APIs
- OpenAI Go SDK 3.64.0 adds agent safety and session-environment controls
- Qwen Code 0.24.2 adds a Linux sandbox foundation and durable workflow controls