crewai - 💡(How to fix) Fix Add --verbose flag for per-step agent execution tracing [2 pull requests]

name: Feature request about: Suggest an idea for this project title: "[FEATURE] " labels: "" assignees: ""

What problem does this solve?

Debugging complex multi-agent workflows is difficult without visibility into individual agent reasoning steps. Currently, developers must add manual print statements or integrate external observability tools to trace agent behavior.

Proposed solution:

Add an optional verbose parameter to Crew initialization (e.g., Crew(verbose=True)) that emits structured logs for each agent execution step:

• Which agent is acting
• Which task it received
• Tool calls with sanitized input/output
• Handoff events between agents

Example output:

[Crew] Starting crew execution
[Agent: researcher] Starting task: research_task
[Agent: researcher] Calling SearchTools.search(query="AI agents")
[Agent: researcher] Task complete, passing to: writer
[Agent: writer] Received context from researcher

Use case:

Developers debugging unexpected agent loops, tracing why a specific tool was called, or teaching team members how agent coordination works without adding print statements manually.