plan-writing

The Plan Writing skill provides a rigorous framework for AI agents to decompose high-level development goals into granular, manageable tasks. Designed for software engineering workflows, this skill moves beyond simple to-do lists by enforcing logical dependency tracking, clear verification criteria, and incremental progress management. It is primarily intended for developers and project managers using AI-assisted coding to maintain high code quality and velocity during complex feature implementation, bug resolution, or system refactoring.

• Enables systematic task breakdown into 2-5 minute actionable units.

• Incorporates mandatory verification criteria for every task, ensuring expected outputs are validated against defined success states.

• Supports dependency mapping and critical path identification to optimize developer throughput.

• Implements dynamic file naming conventions (task-slug.md) to maintain project organization within the root directory.

• Discourages rigid, one-size-fits-all templates, favoring unique, task-specific strategies tailored to the project stack.

• When initiating a new project, prioritize stack decision, MVP scoping, and folder structuring as initial tasks.

• For feature additions, clearly identify affected files, necessary external dependencies, and specific verification testing methods.

• For debugging, document root cause analysis, specific lines of code to modify, and post-fix validation procedures.

• Limit plans to 10 tasks; if a project requires more, partition the plan into multiple files to maintain focus and clarity.

• Always keep verification as the final phase in every task cycle to ensure system integrity.

• Avoid unnecessary verbosity; ensure each step is specific, such as providing exact shell commands or file paths instead of generic instructions like 'setup project' or 'test API'.