writing-plans

Generates structured TDD implementation plans with exact file paths and task persistence.

1|Updated Feb 10, 2026
One-click install
npx skills add https://github.com/stefanfaur/roach-marketplace --skill writing-plans-stefanfaur
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-plans
Source: https://github.com/stefanfaur/roach-marketplace/tree/main/roach/skills/writing-plans
Command: npx skills add https://github.com/stefanfaur/roach-marketplace --skill writing-plans-stefanfaur

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a spec or feature idea into an actionable, step-by-step implementation plan is error-prone: plans often contain vague placeholders, invented file paths, or inconsistent interfaces between tasks. This Skill produces rigorous, codebase-grounded implementation plans that any executor (human or agent) can follow without guessing. ## Core Features & Use Cases - Codebase-grounded planning: Explores the actual repository to verify exact file paths, existing patterns, and test conventions before writing any task. - TDD task structure: Every task follows a failing-test-first cycle with complete code, exact commands, expected outputs, and commit steps. - Interface contracts between tasks: Each task declares what it consumes and produces, so implementers never guess names or signatures from neighboring tasks. - Task persistence: Saves a .tasks.json file alongside the plan so any new session can resume execution from where it left off. - Use Case: You have a spec for a new payments feature. Invoke the skill with the domain, feature description, and constraints; it returns a dated plan file under thoughts/shared/plans/payments/ ready for subagent-driven or parallel-session execution. ## Quick Start Ask the AI to use the writing-plans skill to create an implementation plan for your feature, providing the domain, a short description, key files, and any constraints.

Frequently Asked Questions about writing-plans

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create an implementation plan from a feature spec?▼

Invoke the writing-plans skill with the domain, feature description, key files, and constraints. It explores the codebase, then writes a dated plan to thoughts/shared/plans/<domain>/ with TDD tasks containing exact paths, code, and commands.

What makes a good implementation plan for AI agents?▼

A good plan uses exact verified file paths, complete code in every step, explicit interface contracts between tasks, and no placeholders like TODO or 'add validation'. Each task ends with one independently testable deliverable and a commit step.

Can I resume plan execution in a new session?▼

Yes. The skill writes a .tasks.json file next to the plan recording each task's status and dependencies. A new session running the executing-plans skill reads this file and continues from the first unfinished task.

When should a spec be split into multiple plans?▼

Split when the spec covers multiple independent subsystems. Each plan should produce working, testable software on its own, so one plan per subsystem keeps tasks right-sized and independently reviewable.

Why does the skill forbid EnterPlanMode during planning?▼

EnterPlanMode traps the session in a mode where Write and Edit are restricted, preventing the plan file from being saved. The skill operates in normal mode so it can write the plan and tasks.json to disk.