How to use the Breakdown Plan agent skill (breakdown-plan)

What is an agent skill and what does breakdown-plan make the agent do?

An agent skill is a folder containing a SKILL.md file. The file holds instructions that an AI agent loads when a request matches the skill's description. The breakdown-plan skill is an issue planning and automation prompt. Its instructions ask the agent to act as a senior project manager with DevOps experience and to turn a feature's artifacts into a GitHub project plan.

The skill expects a complete set of feature documents as input: a feature PRD, a technical breakdown and an implementation plan. From these, the agent is asked to produce two files: a project-plan.md and an issues-checklist.md, both stored under a docs/ways-of-work/plan path for the epic and feature.

• Project overview: feature summary, success criteria, milestones without timelines, and risks.
• Work item hierarchy: Epic, Feature, Story, Enabler, Test and Task, shown as a Mermaid diagram.
• GitHub issue templates for Epic, Feature, User Story and Technical Enabler, each with labels, dependencies and estimates.
• A priority and value matrix mapping P0 to P3 levels to labels such as priority-critical and value-high.

Install the breakdown-plan skill

Download the ZIP file for the skill from this page. You can install it in one of two ways.

For Claude Code, unzip the file into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only in that repository. The skill folder should contain SKILL.md.

In the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.

• Personal install: ~/.claude/skills/ (available across your projects).
• Project install: .claude/skills/ inside the repository (shared with anyone who works in that project).

Prepare your feature documents before you ask

The skill does not invent a project from nothing. The prompt expects three files at specific paths, using your own epic and feature names in place of the placeholders. If these files are missing, the output will be generic or made up.

Related prompts in the same collection are plan-test for test strategy, plan-epic-arch for architecture and plan-feature-prd for feature requirements. They are optional, but the source recommends them if you lack the artifacts.

• Feature PRD: docs/ways-of-work/plan/{epic-name}/{feature-name}.md
• Technical breakdown: docs/ways-of-work/plan/{epic-name}/{feature-name}/technical-breakdown.md
• Implementation plan: docs/ways-of-work/plan/{epic-name}/{feature-name}/implementation-plan.md

Trigger the skill with a natural request

Agents decide when to load a skill by matching your request to its description. Name the feature, point to its documents and say what you want produced. Mention GitHub, because the templates are GitHub-specific.

A good request names the epic and feature, confirms the three documents exist, and asks for the two deliverables. Vague requests such as 'make me issues' give the agent less to work with.

• Example: Using the breakdown-plan skill, create the GitHub project plan for the checkout-redesign feature in the payments epic, based on its PRD, technical breakdown and implementation plan.
• Example: Read the feature docs for the search epic and draft the project-plan.md and issues-checklist.md with dependencies and priorities.
• Example: Turn the existing technical breakdown into Enabler and Task issue templates with t-shirt size estimates.

Review the output before filing issues

The skill produces markdown templates and a checklist. It does not create issues, connect to GitHub or contact any network service. You or your team copy the content into GitHub Issues or Projects yourself. Issue numbers in the templates are placeholders such as #{feature-issue-number} that you fill in as issues are created.

Check that the hierarchy matches your documents, that dependencies are real, and that priorities reflect your team's view. The skill's own assessment is that it is a planning template rather than an automation tool, despite the name 'automated tracking' in its description.

• Confirm every story traces back to a requirement in the PRD.
• Check that each Blocked by and Blocks entry is a real dependency.
• Replace template placeholders and estimates with values your team agrees on.

Safety notes: scripts and network access

According to the skill's listing, breakdown-plan runs no scripts and uses no network. Its only file is SKILL.md. This makes it a low-risk instruction file, but you should still read it before installing and review anything the agent writes into your repository.

Because the skill is plain instructions, its effects depend on what the agent then does with your files and tools in the session. Keep your permissions tight if you are unsure.

• Files: SKILL.md only.
• Scripts: none.
• Network: none.

Tips for better plans

The quality of the plan depends mostly on the quality of the input documents. Specific acceptance criteria and clear user types in the PRD produce better stories. Vague technical documents produce vague enablers.

Use the optional related prompts when a document is missing, rather than asking breakdown-plan to fill the gap from guesswork.

• Write acceptance criteria in testable terms before you run the skill.
• Keep one feature per run so the paths and hierarchy stay clear.
• Ask for test issues separately with plan-test if you need a full test strategy.
• Use the template labels as a starting point and align them with the labels your repository already uses.