Choreography for AI-assisted software development. Defers heavy lifting to specialists like superpowers; provides governance, workflows, and templates that produce maintainable projects.
Analyze AI-facing surfaces (skills, hooks, scripts, agent instruction files) for prompt-injection, instruction-hijacking, and untrusted-data-flow risk. Runs as a fresh-context driver. The AI-surface lane of the B4 review stage.
Interview the user to determine project shape, match to an architecture archetype, surface essential decisions, recommend spikes, and scaffold ARCHITECTURE.md and group documents. Standalone or called by /init-project.
Thin orchestrator for the build stage. Reads S2 input frontmatter from the design spec; dispatches Branch 2 (TDD assessment) and Branch 3 (implementation mode) by reading frontmatter values; writes pipeline-state via WS9's helper; exits with explicit success/escape-hatch status. Use as the build-stage entry point after /design exits.
Post-merge cleanup — verify merge state, safely remove the merged local branch/session worktree, and verify spec status. Use after a PR has been merged.
Reconcile a branch's code with its governance documentation across four levels — architecture, governed specs, file headers + top-of-file comments, and (opt-in) function/inline comments. Regenerates AUTO sections, preserves HUMAN sections (stale-flagging broken refs), appends harvested decisions from design specs and plans.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Arboretum is choreography. Superpowers does the work.
An organizational framework for building software with AI code agents.
Arboretum gives domain experts a repeatable way to create projects that are well-organized, maintainable, and understandable — even when the human didn't write most of the code.
It externalizes the practices that experienced software engineers carry intuitively — project structure, work sequencing, safe evolution — into a process that anyone can follow with an AI code agent.
Domain experts building software with AI code agents. If you have deep knowledge in your field but limited experience structuring software projects, this framework gives you opinionated guidance on how to organize what you're building, how to sequence the work, and how to evolve your project safely.
See PRINCIPLES.md for the full set. The short version:
| Pillar | What it is | Where it lives |
|---|---|---|
| Principles | How software should be organized | PRINCIPLES.md |
| Workflows | Best-practice sequences for common scenarios | workflows/ |
| Skills | Discrete operations (arboretum + external) | .claude/skills/ |
| Templates | Standardized formats that skills use | docs/templates/ |
Seven workflows cover the full development lifecycle:
new-project Start from nothing → architecture → spike-spec cycles → build
feature Planned work → survey → design → plan → build → ship
bug-fix Investigate → classify → fix → ship
explore Spike → learn → decide (build, spike again, or abandon)
publish Prepare your project for public sharing
refactor Restructure without changing behaviour
documentation Docs-only changes
See workflows/README.md for detailed stage-by-stage guides.
Arboretum scales with your project through three layers:
CLAUDE.md, specs, core skills. Every project starts here. Simple enough that it feels like "just good organization."Start simple. Add governance only when you feel the pain of not having it.
Arboretum installs as a Claude Code plugin and scaffolds projects via the /arboretum:init skill — no separate git clone of arboretum required.
# In any Claude Code session, install superpowers (arboretum's required peer):
/plugin marketplace add anthropics/claude-plugins-official
/plugin install superpowers@anthropics-claude-plugins-official
# Install arboretum:
/plugin marketplace add stvangaal/arboretum
/plugin install arboretum@arboretum
# Then in any new project directory, scaffold the governed structure:
/arboretum:init
/arboretum:init creates CLAUDE.md, PRINCIPLES.md, docs/, workflows/, hooks, and git config; verifies superpowers is installed; and hands off to /architect for the project-shape interview. The plugin updates from upstream automatically on subsequent /plugin install runs — projects don't carry frozen framework copies.
Permission denied (publickey) during install — Claude Code clones plugins via SSH by default. If you haven't added an SSH key to GitHub, run this once in your terminal to switch to HTTPS instead:
git config --global url."https://github.com/".insteadOf "[email protected]:"
Then retry the /plugin install command.
See examples/rule-flow-engine/ for a fully governed sample with architecture, definitions, specs, register, and version pins.
npx claudepluginhub stephen-van-gaal/arboretum --plugin arboretumCore skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.