YMD/PMD format specification and authoring tools for structured, modular AI prompts
You can install this plugin from any of these themed marketplaces. Choose one, add it as a marketplace, then install the plugin.
Choose your preferred installation method below
A marketplace is a collection of plugins. Every plugin gets an auto-generated marketplace JSON for individual installation, plus inclusion in category and themed collections. Add a marketplace once (step 1), then install any plugin from it (step 2).
One-time setup for access to all plugins
When to use: If you plan to install multiple plugins now or later
Step 1: Add the marketplace (one-time)
/plugin marketplace add https://claudepluginhub.com/marketplaces/all.jsonRun this once to access all plugins
Step 2: Install this plugin
/plugin install gradient-2@allUse this plugin's auto-generated marketplace JSON for individual installation
When to use: If you only want to try this specific plugin
Step 1: Add this plugin's marketplace
/plugin marketplace add https://claudepluginhub.com/marketplaces/plugins/gradient-2.jsonStep 2: Install the plugin
/plugin install gradient-2@gradient-2A proposal for a layered context architecture
Gradient defines architectural patterns for building layered context injection systems for Claude Code plugins, emphasizing smooth transitions from normative specifications through applied context to dynamic orchestration.
███▓▓▒▒░░░░ SPECS (The WHAT - Normative)
███▓▓▒▒░░ CONTEXT (The HOW - Applied)
██▓▓▒▒ PROMPTS (The ACTION - Orchestration)
Gradient is an architectural framework for organizing Claude Code plugins and context injection systems. It emerged from patterns discovered while building projects like ymd-spec, semantic-docstrings, and code-zen.
Instead of rigid, isolated layers, Gradient promotes smooth transitions between architectural layers:
Like a gradient, each layer dissolves into the next, creating a fluid, organic architecture.
@ references for specsgradient/
├── gradient/spec/ # SPECS: Normative (The WHAT)
│ ├── architecture-spec.md
│ ├── anti-duplication-principles.md
│ └── layer-spec.md
│
├── context/ # CONTEXT: Applied (The HOW)
│ ├── examples.md
│ ├── implementation-guide.md
│ └── decision-guide.md
│
├── prompts/ # PROMPTS: Orchestration (The ACTION)
│ └── load-context.md # Thin loader with @ references
│
├── commands/ # COMMANDS: Entry points (API)
│ └── *.md # One-to-one with prompts
│
├── agents/ # AGENTS: Specialized contexts
│ └── *.md # Spin-off processes
│
└── docs/ # DOCS: Human documentation
└── *.md # Architecture guides, tutorials
SPECS (Normative):
CONTEXT (Applied):
PROMPTS (Orchestration):
@ referencesLayers blend into each other:
Like REST routes, commands are thin entry points:
<!-- commands/load-gradient-context.md -->
@~/.claude/gradient/prompts/load-context.md
One-to-one mapping with prompts. Commands define the interface, prompts orchestrate the logic.
Specialized contexts with separate permissions:
<!-- agents/architecture-reviewer.md -->
You are an architecture reviewer for Gradient projects.
@~/.claude/gradient/gradient/spec/architecture-spec.md
@~/.claude/gradient/context/decision-guide.md
Run independently, return summarized results without contaminating main context.
Triggered by Claude Code lifecycle events (not yet explored in this project).
When code is more efficient than tokens:
# scripts/validate-structure.sh
# Validate Gradient architecture compliance
Documentation Site: Visit the complete documentation site (when published) with interactive examples and guides.
Local development: Run the Jekyll site locally:
cd docs
bundle exec jekyll serve
# Access at http://localhost:4000
Quick reads:
Load the architecture spec:
@~/.claude/gradient/gradient/spec/architecture-spec.md
@~/.claude/gradient/gradient/spec/anti-duplication-principles.md
Status: In Development
This project is actively being developed as a distillation of patterns from:
ymd-spec: YMD/PMD format specificationsemantic-docstrings: Semantic documentation standardscode-zen: Zen of Python implementation guide"Architecture isn't about rigid layers—it's about smooth, intentional transitions."
Gradient recognizes that effective context injection requires:
Like a gradient in design, our architecture dissolves boundaries while maintaining distinct identities.
Complete documentation with interactive examples, Mermaid diagrams, and step-by-step guides:
MIT License - See LICENSE file for details
Contributions welcome! This project aims to establish patterns for Claude Code plugin architecture. Share your insights and patterns.
Gradient - From specs to action, smoothly.
0.1.0