AI Resources Lifecycle Platform
Design, develop, test, and deliver AI agent skills, rules, and workflows.
Fork this. Build resources. Deliver via Forge.
Convention over configuration
Put a file in the right place — the platform discovers it. No config, no registration.
| What | Convention | Discovered by |
|---|---|---|
| Skill | skills/<name>/SKILL.md | /create-resource, /test-resource, Forge |
| Rule | rules/<name>.md | /create-resource, Forge |
| Workflow | workflows/<name>.md | /create-resource, Forge |
| Specimen | orphan branch specimen/<lang>/<fw> | /test-resource |
The full lifecycle
From idea to delivery — the platform guides every step.
One orchestrator, three phases
/design-resource
Creates design.md and architecture diagrams for the new resource.
/develop-resource
Scaffolds and implements the resource from the design.
/test-resource
Type-aware testing against real project specimens.
/create-resource orchestrates all three phases end-to-end.
Included resources
Skills
| Skill | Version | Type | Description |
|---|---|---|---|
| /drawio | 1.0.0 | utility | Generate draw.io diagrams as native files with PNG/SVG/PDF export |
| /detect-stack | 1.0.0 | utility | Classify project type, language, framework, and build tools |
| /review-pr | 1.0.0 | orchestrator | Review PR changes for quality, style, and correctness |
| /document | 1.0.0 | orchestrator | Generate project documentation with architecture diagrams |
Rules & Workflows
drawio-conventions
Standards for clean, readable draw.io diagrams
code-style
General coding conventions and best practices
documentation-standards
Documentation quality and completeness requirements
document-project
Workflow: end-to-end documentation using /document skill
Test against real projects
Specimens are real project snapshots on orphan branches. Any skill that operates on source code can use them.
specimen/python/fastapiFastAPI REST API service snapshot
specimen/ts/expressExpress.js TypeScript REST API snapshot
Delta branches add targeted changes with intentional findings for testing edge cases.
Flexible delivery. Use what works for you.
The platform is an abstract idea that structures your AI resources and helps you manage them. That's it. How you deliver them to developers is entirely up to you.
Forge CLI
Recommended. Set-and-forget sync with auto-transforms for every AI tool. The easiest path.
Git submodules
Classic approach. Add the platform repo as a submodule and symlink what you need.
Custom tooling
Already have a delivery mechanism? The platform's directory conventions work with any tool that can read a git repo.
Best with Forge
Fork the starter, build your team's skills, then deliver to every developer with Forge. Configure once, forget it exists.