project-plan

Creates design documents and milestone-based implementation plans with testing, review, and handoff gates.

2|Updated May 13, 2026
One-click install
npx skills add https://github.com/curtisgalloway/public-skills --skill project-plan-curtisgalloway
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: project-plan
Source: https://github.com/curtisgalloway/public-skills/tree/main/plugins/agent-workflow/skills/project-plan
Command: npx skills add https://github.com/curtisgalloway/public-skills --skill project-plan-curtisgalloway

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running software projects lose context between agent sessions, causing repeated work, skipped reviews, and unfinished milestones. This Skill produces durable design and implementation plan documents so any fresh session can resume work without reconstructing decisions from chat history. ## Core Features & Use Cases - Design-first workflow: Writes or amends an authoritative design document with requirements, architecture, and acceptance criteria, then pauses for explicit user approval before planning. - Milestone-based implementation plan: Derives session-sized milestones with dependencies, acceptance criteria, verification commands, review methods, and evidence files, plus a final system-wide verification gate. - Enforced execution gates: Requires tests, an artifact-producing review, and a checkpoint commit before any milestone is marked complete, and writes a structured handoff for the next session. - Use Case: Ask the agent to plan a multi-week feature; it writes docs/design.md, gets your approval, derives docs/implementation-plan.md with milestones M1-M5, then executes one milestone per session with recorded evidence. ## Quick Start Ask the agent to turn this feature idea into a design document and milestone-based implementation plan for the project.

Frequently Asked Questions about project-plan

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

FAQPage Schema
How do I break a large software project into milestones with an AI agent?▼

Start with a design document defining requirements and acceptance criteria, then derive milestones ordered by dependencies and risk. Each milestone should deliver one coherent, testable capability sized to fit a single session including implementation, tests, review, and handoff.

How to resume an agent coding session without losing project context?▼

Keep the design, implementation plan, and a handoff section in versioned documents rather than chat history. A new session reads the design, plan status, latest evidence file, and handoff to identify the current milestone and exact next action.

Does this planning workflow require a specific AI agent or tool?▼

No, it uses standard file reading, editing, and search tools available in any agent harness. Executing milestones additionally requires the project's own build and test tools, which the plan identifies up front.

Why does the skill pause after writing the design document?▼

Every milestone inherits flaws in the design, so user approval at that point is the cheapest place to fix errors. The pause can be waived explicitly, and the waiver is recorded in the plan's conventions block for later sessions.

What happens if a milestone cannot finish in one session?▼

The milestone stays marked in_progress or blocked with an incomplete handoff explaining remaining work, and the next session resumes it rather than advancing. Milestones that clearly cannot fit should be split before starting, preserving testable boundaries.