Native Claude Code Skill

Think beforeyou write code

In the AI era, coding is no longer the bottleneck — decisions are. PM-Quill turns vague ideas into clear product blueprints, so every line of code counts.

PM-Quill — Claude Code
$ /spec # describe your idea
Product spec generatedspecs/my-app/spec.md
$ /feasibility # evaluate feasibility
3 technical approaches comparedspecs/my-app/feasibility.md
$ /plan # break down the plan
12 executable tasks generatedspecs/my-app/plan.md
🎉 Ready to hand off to Cursor / Claude Code
⚠ THE PROBLEM

AI writes your code
But who helps you decide what to build?

Cursor and Claude Code made coding 10x faster. But if the direction is wrong, you're just building the wrong thing faster.

  • 💡

    "Code first, think later"

    Start coding with a vague idea, realize halfway through the requirements were never clear

  • 🔄

    Constant pivots

    Three prototype versions later, always thinking "I should have done it differently..."

  • 👤

    Solo product manager

    Indie developers without a PM to challenge their assumptions, relying on gut feeling alone

Abandoned

AI Content Aggregator v3

2 weeks of coding, then discovered 10 competitors already exist

Pivoted

TODO App → Project Mgmt → ???

Features keep growing, core value keeps fading

Stuck

Can't decide what to cut from MVP

"This is important too" "Can't skip that either"

Abandoned

Social Commerce Mini-App

Changed tech stack 3 times, eventually gave up

Four steps, from idea to action plan

Each step powered by a specialized Agent, outputting structured Markdown files — trackable and version-controlled

01
/spec

Product Spec

Input a vague idea, output a structured Spec: target users, core features (max 3), out of scope, success criteria

spec.md generated
User persona defined
3 core features
5 "out of scope" items
02
/feasibility

Feasibility Analysis

Evaluate technical feasibility with 2-3 approach comparisons, effort estimates, and key risks identified

feasibility.md generated
Plan A vs B vs C
Risk: 3rd-party API limits
Estimate: 2-week MVP
03
/plan

Execution Plan

Break confirmed Spec into executable tasks, each with acceptance criteria, ready to hand off to coding AI

plan.md generated
12 tasks prioritized
MVP scope defined
Ready for Cursor
04
/review

Project Review

Post-project retrospective: what decisions were right, what went wrong, lessons learned for the next project

review.md generated
3 key takeaways
2 improvements needed
Archived to archive/

Why PM-Quill

Not another AI coding tool — a thinking partner before you code

🎯

Spec-driven, fewer detours

Clarify what to build and what NOT to build before writing the first line of code. A good Spec beats ten refactors.

Progressive, no fluff

Won't bombard you with 100 questions upfront. Delivers an 80% solution first, flags uncertainties for your decision.

📄

Files as database

All outputs are Markdown files. Readable, searchable, Git-versioned. No platform lock-in.

🔗

Seamless AI handoff

Output tasks can be directly copied to Cursor or Claude Code. PM-Quill decides the direction, they write the code.

📊

Decisions on record

Every product decision is documented: why this approach, what was dropped, where the risks are. No guessing six months later.

🛠

Zero config, ready to go

Runs as a native Claude Code skill. No extra installation. Open terminal, type a command, start building.

The difference PM-Quill makes

❌ Without PM-Quill

  • Jump into coding the moment an idea hits
  • Requirements live in your head, changing on the fly
  • Pick tech stack by gut feeling, regret later
  • MVP tries to include everything, never ships
  • After launch, no idea what to prioritize next
  • Three months later, can't remember why anything was designed that way

✔ With PM-Quill

  • Structure ideas first with /spec
  • Requirements documented, version-tracked
  • 2-3 approaches compared with evidence
  • MVP scope clear — only the 3 most critical things
  • Prioritized task list, ready to execute
  • All decisions on record, /review captures lessons

Upstream of coding AI

Cursor / Claude Code writes your code. PM-Quill decides what code to write.

💡 Ideation

PM-Quill

/spec → /feasibility → /plan

💻 Coding

Cursor / Claude Code

Execute coding based on PM-Quill's task list

📦 Management

Linear / Notion

Project management, progress tracking, team collaboration

🎨 Design

Figma / v0

UI design, prototyping, visual mockups

You might be a PM-Quill user

Core User

Indie Tech Founder

Can code, but never been a product manager

  • "I want to build an AI-powered XX" — don't know where to start
  • Found an interesting open-source project, want to build a product on it, but unsure about direction
  • MVP is done, but what's the next priority?
  • Solo full-stack, need an AI product manager to sanity-check decisions
Potential User

Vibe Coder

Non-technical background, building products with AI coding tools

  • Can't write code at all, but has product intuition
  • AI can write code, but nobody helps think through what to build
  • Needs a structured thinking framework, not just a chatbot
  • Wants every decision documented for team alignment

Start with PM-Quill
Think first, build right

Open your terminal, type one command, let AI do what a product manager should do.

$ /spec I want to build a...


⭐ Get PM-Quill on GitHub

Native Claude Code Skill · Zero Config · Ready to Use