CrewAI 1.15.33
CrewAI 1.15.33 introduces CLI skill search, input passing to Python flows, deploy-by-zip functionality, and fixes token reporting for LiteLLM tool calls and structured outputs.
Why this signal matters
CrewAI release 1.15.33 introduces multiple CLI enhancements and platform bug fixes. Key features include the crewai skill search command to list organization skills, the ability to pass inputs directly to Python flows via crewai run --inputs , and a deploy create --zip command that prints the associated build commit. Fixes include accurate LiteLLM token reporting for tool calls and structured outputs, backwards-compatible import restoration, backwards compatibility between the JSON crew runner and older crewai-core versions, and persisting checkpoint callbacks gracefully without breaking restores. Documentation updates cover agent planning via PlanningConfig and a migration of the reasoning page.
Actionable summary
CrewAI v1.15.33 adds crewai skill search , --inputs support for flows in crewai run , and deploy create --zip . Critical bug fixes address LiteLLM token usage tracking, JSON crew runner compatibility across core versions, flow checkpoint restoration, and CLI/TUI stability.
- Agent usefulness
- 62/100
- Confidence
- 96%
- Canonical data
- JSON + Markdown
What builders should check
- Upgrade to CrewAI 1.15.33 using your project package manager.
- Adopt crewai run --inputs if passing dynamic input arguments directly to Python flows from the CLI.
- Review agent planning configurations against the updated PlanningConfig documentation.