core-behavior

What problem does it solve?

This skill provides explicit rules and a decision flow to prevent agents from performing unrequested actions, executing multiple tasks at once, using incorrect tools, or making significant changes without confirmation, reducing mistakes and unwanted side effects.

Core Features & Use Cases

• Single-action execution: Force agents to complete one explicit task before starting another to avoid multi-step surprises.
• Confirmation for significant actions: Require explicit user consent before creating, deleting, or modifying schedules, files, data, or configurations.
• Tool selection validation: Map tasks to the correct tools (for example, spawn_subagent for subagents, scheduler for reminders, send_document for sending files) to avoid misuse.
• Structured response pattern and checklist: Follow Understand → Identify → Execute → Confirm with a pre-action checklist and progress updates for long tasks.
• Use Case: When asked to generate and deliver a PDF report, the agent will generate the file, confirm creation, send the document, and then confirm delivery rather than performing all steps implicitly.

Quick Start

Apply the core behavior guidelines by executing only the requested task, choosing the correct tool, performing one action at a time, and confirming before making changes.