Grimoire domain decomposition skills for Claude Code
npx claudepluginhub tapania/grimoire-skillsAI-powered domain decomposition pipeline — transform a project vision into governance-checked, dependency-resolved implementation slices through structured exploration
A suite of Claude Code skills that implement the Grimoire methodology — an AI-powered domain decomposition pipeline that transforms a project vision into governance-checked, dependency-resolved implementation slices.
For the full narrative of how these skills work together, read VISION.md.
Grimoire takes a software project from "here's what I want to build" to "here's a structured, validated specification tree with delivery slices." The pipeline:
The pipeline is sequential but not mandatory — stages can be skipped. All state is stored as YAML files in a .grimoire/ directory. No server, no database — just files the agent reads and writes.
| Skill | Command | Description |
|---|---|---|
| grimoire | /grimoire | The main entry point. Describes the full pipeline, detects current stage, and routes to the right skill. |
1. VISION → 2. GOVERNANCE → 3. TAXONOMY → 4. EXPLORATION → 5. SLICING
| # | Skill | Command | What It Does | Output |
|---|---|---|---|---|
| 1a | grimoire-vision | /grimoire-vision | Structured vision declaration (domain, audience, constraints) | .grimoire/vision.md |
| 1b | grimoire-vision-playground | /grimoire-vision-playground | Creative identity exploration (taxonomy → features → prose narrative) | .grimoire/vision.md, vision-identity.yaml |
| 1c | grimoire-archaeology | /grimoire-archaeology | Bootstrap from existing codebase (reverse-engineer vision, schools, tree) | Full .grimoire/ directory |
| 2a | grimoire-governance | /grimoire-governance | Define schools with sealed/advisory wards + scoring criteria | .grimoire/schools.yaml, scoring.yaml |
| 2b | grimoire-school-tree-composer | /grimoire-school-tree-composer | Compose governance tree from archetype/overlay catalog (advanced) | .grimoire/schools.yaml |
| 3 | grimoire-taxonomy | /grimoire-taxonomy | Decompose domain into specification tree (alternatives → expand) | .grimoire/tree.yaml |
| 4a | grimoire-expand | /grimoire-expand | Expand a branch with child capabilities | Updates tree.yaml |
| 4b | grimoire-ritual | /grimoire-ritual | Full exploration loop: lenses → suggest → gate → tag → score | Ranked suggestions → tree.yaml |
| 5 | grimoire-slicing | /grimoire-slicing | Propose dependency-complete delivery slices with validation | .grimoire/slices/*.yaml |
| Skill | Command | Description |
|---|---|---|
| grimoire-gate-check | /grimoire-gate-check | Enforce sealed wards and structural gates (sole blocking authority) |
| grimoire-familiar | /grimoire-familiar | Analyze a dependency's impact on the tree |
| grimoire-pathfinder | /grimoire-pathfinder | Critical path and bottleneck analysis |
| grimoire-refactor | /grimoire-refactor | Tree structural surgery (merge, split, reparent, prune) |
| grimoire-retrospective | /grimoire-retrospective | Post-implementation calibration against actuals |
# Add the marketplace
/plugin marketplace add tapania/grimoire-skills
# Install the plugin
/plugin install grimoire-skills@tapania-grimoire-skills
Or use the interactive plugin manager:
/plugin
Navigate to Marketplaces tab, add tapania/grimoire-skills, and install from the Discover tab.
Requirements: Claude Code version 1.0.33 or later (claude --version to check)
git clone https://github.com/tapania/grimoire-skills.git
cp -r grimoire-skills/skills/ your-project/.claude/skills/
git clone https://github.com/tapania/grimoire-skills.git
cp -r grimoire-skills/skills/* ~/.claude/skills/
For a new project, run skills sequentially:
/grimoire-vision → /grimoire-governance → /grimoire-taxonomy → /grimoire-expand → /grimoire-ritual → /grimoire-slicing
Or just run /grimoire and it will guide you through the pipeline based on current state.
Matt Pocock's skills for real engineering, as an installable Claude Code plugin.
Plugins by Alex Newman (thedotmack)
Ultra-compressed communication mode for Claude Code. Cuts 65% of output tokens (measured) while keeping full technical accuracy.