crewAI 1.15.29 adds evaluation execution and export
- crewAI now accepts kwargs that are compatible with LiteLLM.
- The CLI can search for applications on the platform.
- AgentExecutor is no longer marked as experimental.
- A single crewAI eval command can run the crew to produce an evaluation verdict.
- Evaluation results can now be exported as files through crewAI eval.
Our take
These evaluation changes can reduce the glue code needed to run a crew and preserve its results in a team’s own workflow. The release does not specify export formats or how the command fits into automated pipelines, so teams should check those details before replacing an existing evaluation harness.
Sources
- crewAI 1.15.29 · crewAI Releases
The summary and analysis were generated by AI. The original sources are authoritative. AI content policy