OpenClaw Ideation Skill

Transform raw brain dumps into structured implementation artifacts through a confidence-gated workflow.

Inspired by nicknisi/claude-plugins/ideation, adapted for OpenClaw.

What It Does

Turn messy, stream-of-consciousness ideas into:

• Contracts — Problem statements, goals, success criteria, scope boundaries
• PRDs (optional) — Phased requirements documents
• Implementation Specs — Detailed technical specs ready for execution

Installation

Option 1: Clone + symlink (recommended)

mkdir -p ~/.openclaw/skills ~/.openclaw/skills-src
git clone https://github.com/SharkJets/openclaw-skill-ideation.git ~/.openclaw/skills-src/openclaw-skill-ideation
ln -s ~/.openclaw/skills-src/openclaw-skill-ideation/ideation ~/.openclaw/skills/ideation

Option 2: degit (no git history)

mkdir -p ~/.openclaw/skills
npx degit SharkJets/openclaw-skill-ideation/ideation ~/.openclaw/skills/ideation

Option 3: Manual clone

mkdir -p ~/.openclaw/skills
git clone https://github.com/SharkJets/openclaw-skill-ideation.git /tmp/oc-ideation
cp -r /tmp/oc-ideation/ideation ~/.openclaw/skills/
rm -rf /tmp/oc-ideation

Option 4: Workspace-local

Clone or copy the ideation/ folder into your project's skills/ directory for per-project use.

After Installation

OpenClaw snapshots skills when a session starts. After installing, start a new session for the skill to be available. A gateway restart alone isn't enough — you need a fresh session.

Alternatively, enable the skills watcher for hot-reload in ~/.openclaw/openclaw.json:

{
  "skills": {
    "load": {
      "watch": true
    }
  }
}

Usage

Just dump your messy ideas:

I want to build a thing that lets users save bookmarks but like, 
not just URLs, also notes and stuff. Maybe tags? Folders feel 
old school. Should work offline I think. Oh and it needs to 
sync across devices somehow...

The skill will:

1. Intake — Accept your mess without judgment
2. Explore — Understand your existing codebase (if applicable)
3. Score — Rate clarity across 5 dimensions (0-100)
4. Clarify — Ask targeted questions until confidence ≥ 95
5. Contract — Generate contract.md for your approval
6. Spec — Create implementation specs with validation commands

Confidence Scoring

Each dimension is scored 0-20 points:

Dimension	Question
Problem Clarity	Do I understand what we're solving and why?
Goal Definition	Are goals specific and measurable?
Success Criteria	Can I write tests for "done"?
Scope Boundaries	Do I know what's in/out?
Consistency	Are there contradictions?

Thresholds:

• < 70: Ask 5+ questions
• 70-84: Ask 3-5 questions
• 85-94: Ask 1-2 questions
• ≥ 95: Generate contract

Output Structure

./docs/ideation/{project-name}/
├── contract.md              # Problem, goals, scope (always)
├── prd-phase-1.md           # Requirements (if PRDs chosen)
├── prd-phase-2.md
├── spec-phase-1.md          # Implementation specs (always)
├── spec-phase-2.md
└── spec-template-{name}.md  # Shared template for repeatable phases

Files

ideation/
├── SKILL.md                    # Main skill definition
└── references/
    ├── confidence-rubric.md    # Detailed scoring criteria
    ├── contract-template.md    # Contract output template
    ├── prd-template.md         # PRD output template
    └── spec-template.md        # Spec output template

Credits

Based on the ideation plugin by Nick Nisi from claude-plugins.

License

MIT