Skip to content

Latest commit

 

History

History
255 lines (176 loc) · 7.95 KB

File metadata and controls

255 lines (176 loc) · 7.95 KB

📐 Skill Format Guide

The complete anatomy of a PM Skills file — every field, every section, explained.


Overview

Every skill in this system is a single SKILL.md file that follows a consistent structure. This consistency is what makes skills portable across AI tools and reliable in output quality.

SKILL.md = YAML Frontmatter + 7 Markdown Sections

Part 1: YAML Frontmatter

The frontmatter is metadata at the top of every skill file, enclosed in --- delimiters.

---
name: problem-statement
description: Frame a customer problem with evidence before jumping to solutions.
type: component
domain: discovery
difficulty: beginner
estimated_time: "15-25 min"
prerequisites: []
outputs: ["problem-statement-document"]
triggers:
  - "I need to define the problem we're solving"
  - "Help me frame the customer problem"
  - "Write a problem statement"
---

Field Reference

Field Required Values Description
name kebab-case string Unique identifier (matches the folder name)
description ≤200 chars One-line summary of what the skill does
type component | interactive | workflow Determines the skill's behavior pattern
domain One of 10 domains Which PM lifecycle stage this belongs to
difficulty beginner | intermediate | advanced Complexity level for the user
estimated_time String (e.g., "15-25 min") How long a typical session takes
prerequisites Array of skill names Skills that should be completed first
outputs Array of artifact names What this skill produces
triggers Array of phrases Natural language phrases that activate this skill

The Three Skill Types

Type Symbol Behavior Count
Component 🧱 Generates a single artifact with fill-in templates. No back-and-forth. 62
Interactive 🔄 Asks clarifying questions first, then generates adaptive output. Multi-turn. 32
Workflow 🎭 Orchestrates multiple skills across phases. Announces transitions. 7

The 10 Domains

Domain Slug Lifecycle Stage
Discovery & Research discovery Understanding the problem space
Strategy & Vision strategy Setting direction and goals
Planning & Prioritization planning Deciding what to build and when
Specification & Design specification Defining what to build in detail
Execution & Delivery execution Building and shipping
Launch & GTM launch Bringing to market
Growth & Optimization growth Growing and improving
Analytics & Metrics analytics Measuring and learning
Communication & Stakeholders communication Aligning people
Leadership & Career leadership Leading and growing as a PM

Difficulty Levels

Level Target Audience Example
Beginner New PMs, first-time framework users Problem Statement, User Story, RACI
Intermediate Experienced PMs, familiar with frameworks PRD, OKRs, Prioritization Advisor
Advanced Senior/Staff PMs, complex multi-dimensional work Strategy Session, Growth Model, VP Readiness

Part 2: Markdown Sections

Every skill file has 7 standard sections after the frontmatter. Here's what each does and why it matters.


Section 1: Purpose

## Purpose

Frame a customer problem with evidence before jumping to solutions. This skill 
forces clarity on who is affected, what is broken, why it matters, and how big 
the impact is.

This is not a solution brief. It's the foundation that every PRD, epic, and 
user story should trace back to.

What it does: 1-3 sentences explaining what this skill produces, when to use it, and what it's not.

Why it matters: Helps the AI (and you) quickly decide if this is the right skill for the job.


Section 2: Key Concepts

## Key Concepts

### The 5W1H Problem Frame
[Core framework explained with tables, diagrams, etc.]

### Why This Works
[Bullet list of reasons this approach is effective]

### Anti-Patterns (What This Is NOT)
[Common misunderstandings to prevent wrong usage]

### When to Use This / When NOT to Use This
[Situational guidance]

What it does: Teaches the core framework, mental model, or methodology behind the skill.

Why it matters: The AI uses this section to ground its output in established PM practice rather than generic advice.


Section 3: Application

## Application

### Step 1: [Action Name]
[Instructions with context]

\```markdown
**Field:** [Template to fill in]
\```

**Quality check:**
- [ ] [Verification question]

What it does: Step-by-step instructions with fill-in templates and quality checks at each step.

Why it matters: This is the core engine of the skill. The AI generates output by following these steps and filling in the templates with your specific context.


Section 4: Examples

## Examples

### ✅ Good Example
[Complete filled-in example]
**Why this works:** [Explanation]

### ❌ Bad Example
[Complete bad example]
**Why this fails:** [Explanation]

What it does: Concrete good and bad examples that calibrate the AI's output quality.

Why it matters: Examples are the most powerful teacher. The AI uses these to understand the quality bar.


Section 5: Quality Rubric

## Quality Rubric

Score each dimension 0-2. Target: 8+ out of 10.

| Dimension | 0 (Missing) | 1 (Partial) | 2 (Excellent) |
|-----------|-------------|-------------|----------------|
| Specificity | ... | ... | ... |
| Evidence | ... | ... | ... |

What it does: 5-dimension scoring system (0-2 per dimension) that the AI uses to self-evaluate its output.

Why it matters: Forces consistent quality. If any dimension scores below 2, the AI should suggest improvements before you finalize.

Scoring guide:

  • 0 (Missing): The dimension is absent or completely inadequate
  • 1 (Partial): Present but incomplete or generic
  • 2 (Excellent): Thorough, specific, and professional-grade

Target: 8+ out of 10 across all 5 dimensions.


Section 6: Common Pitfalls

## Common Pitfalls

### Pitfall 1: [Descriptive Name]
**Symptom:** [What you'll see when this happens]
**Consequence:** [Why it's bad]
**Fix:** [How to avoid or correct it]

What it does: Named failure modes with symptoms, consequences, and fixes.

Why it matters: Proactive pitfall detection. The AI flags these during generation, not after you've shared a flawed artifact with stakeholders.


Section 7: References

## References

### Related Skills
- `skills/domain/skill-name/SKILL.md`[How it relates]

### External Frameworks
- [Framework Name][Brief description and source]

What it does: Cross-references to related skills and external frameworks.

Why it matters: Enables skill discovery and chaining. After completing one skill, the AI can suggest related skills for your next step.


Quick Reference: Section Checklist

Use this when reviewing or creating a skill:

  • Frontmatter — All 9 fields present with valid values
  • Purpose — Clear what/when/why in 1-3 sentences
  • Key Concepts — Core framework + anti-patterns + when/when not to use
  • Application — Step-by-step with templates + quality checks per step
  • Examples — At least 1 good + 1 bad with explanations
  • Quality Rubric — 5 dimensions, 0-2 scale, target 8+/10
  • Common Pitfalls — 2-4 named pitfalls with symptom/consequence/fix
  • References — Related skills + external frameworks

📚 Back to Docs → · ✍️ Build a Skill →