What is the maximum size of a dissected plan file?

Each generated plan file is capped at 50 lines to ensure it remains focused and manageable for an AI agent or developer to execute independently.

How does the skill handle dependencies between tasks?

It utilizes YAML frontmatter with a 'depends-on' key in the generated files to track which tasks must be completed before others can begin.

What happens if a plan section is too large?

The skill automatically splits oversized sections into multiple smaller plans while ensuring all relevant design decisions and constraints are preserved in each file.

Is there a way to verify the dissection is complete?

Yes, the skill includes a dedicated validation phase where a subagent audits the output against the original document to ensure no requirements or context were missed.

Where are the dissected plans stored?

All outputs are organized within a dedicated 'plan-pit/' directory at the project root, structured by the source document name for easy navigation.

Plan Dissector

Name: Plan Dissector
Author: SaintPepsi

bySaintPepsi

0•

Productivity & Workflow

Breaks down complex design documents into small, manageable, and agent-executable task files.

Plan Dissector is a specialized Claude Code skill designed to transform oversized specifications and lengthy design documents into a structured series of independent, executable plan files. By enforcing a strict 50-line limit and requiring self-contained context for each chunk, it ensures that AI agents can implement features without losing critical design rationale or getting overwhelmed by context window limits. The skill automatically manages dependencies, creates a standardized folder hierarchy, and runs a comprehensive validation pass to guarantee that no requirements are lost during the transition from high-level planning to actionable execution.

Key Features

01Standardized 'plan-pit' folder structure for organized project management

02Automated validation reports cross-checking completeness and self-containment

03Automatic decomposition of large documents into <50-line executable chunks

04Intelligent dependency tracking using YAML frontmatter for sequential execution

05Context-aware subagent dispatching for processing massive files without detail loss

060 GitHub stars

Use Cases

01Onboarding AI agents to complex features with modular, step-by-step instructions

02Converting a monolithic PRD or technical spec into an actionable implementation roadmap

03Preparing large-scale refactoring plans for autonomous agent execution

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saintpepsi/brainstorm-pipeline dissect-plan

For use in Claude.ai and ChatGPT

Download Skill