make_plan

To coordinate multi-step tasks across modules, you can query a stronger agent for a concrete, executable plan and execute it locally step by step, ensuring clear dependencies and checkpoints are maintained throughout the workflow.

Generating a detailed execution plan involves delegating the planning phase to a more capable agent, which produces a step-by-step strategy with clear dependencies that you can then execute and track within your local environment.

Yes, task delegation requires the listAvailableAgents and delegateToAgent tools to discover capable agents and request plans, alongside the skillFileTool to persist the generated plan in your workspace for traceability.

Yes, you can save the multi-step workflow plan to your workspace using the skillFileTool, which preserves the detailed plan for future reference and ensures full traceability of the execution steps and dependencies.

To refine actionable steps lacking specificity, you can use targeted prompts to request more detailed instructions from the stronger agent, ensuring each step in the multi-step plan is concrete enough for local execution.