Skip to content
Skill

project-spec

by Ilm-Alan

AI Summary

This booster guides AI agents through a structured process to create comprehensive project specifications autonomously, enabling them to understand scope, architecture, and constraints before implementation. Developers and AI systems benefit by having clear, actionable project blueprints that facilitate autonomous work and reduce ambiguity.

Install

Copy this and paste it into Claude Code, Cursor, or any AI assistant:

I want to install the "project-spec" skill in my project.

Please run this command in my terminal:
# Install skill into the correct directory
mkdir -p .claude/skills/project-spec && curl --retry 3 --retry-delay 2 --retry-all-errors -o .claude/skills/project-spec/SKILL.md "https://raw.githubusercontent.com/Ilm-Alan/project-spec/main/SKILL.md"

Then restart Claude Code (or reload the window in Cursor) so the skill is picked up.

Description

Create a project spec for AI agents to work from autonomously.

Project Spec

What to build, how it fits together, how to keep going.

Process

• Explore: Check context—files, docs, commits, patterns. • Ask: Ask many questions (multiple choice when possible) until you can fill every section. • Approaches: If multiple valid paths exist, present 2-3 options. Lead with your recommendation. • Derive: Map conversation to spec sections: • Mission: what, who, why • Architecture: technical decisions • Components: what exists • Constraints: non-negotiables • Priorities: current focus • Standards: quality bar • Write: Output PLAN.md with spec and progress in one file.

Spec Structure (under 100 lines)

Instructions: • Read Progress section to know what's done, check git log for context • Build the next item from Priorities—do not ask what to work on • Make decisions from existing code patterns • Commit at natural breakpoints (test passing, working feature, integration complete) • Update Progress section after each commit, continue to next item • When priorities clear, surface new work • PLAN.md is a living document—update as work evolves • No clarifying questions—ship working features Mission: 1-3 lines. What, who, why. Architecture: Technical decisions stated, not debated. Components: What exists. One line each—name and essence. Constraints: Non-negotiables only. Priorities: Current focus. Replenishes. Standards: Quality bar. Improve over time. Progress: What's done, what's current, what's next. Update after each commit.

Writing Principles

Concrete, not vague • "CLI tool for rotating AWS credentials" not "developer utility" • "Postgres, Redis, S3" not "appropriate storage solutions" No filler • Cut: "Prefer working software over perfection", "Start simple", "Future integrations"

Discussion

0/2000
Loading comments...

Health Signals

MaintenanceCommitted 2mo ago
Active
AdoptionUnder 100 stars
1 ★ · Niche
DocsMissing or thin
Undocumented

GitHub Signals

Stars1
Issues0
Updated2mo ago
View on GitHub
MIT License

My Fox Den

Community Rating

Sign in to rate this booster

Works With

Claude Code