copilot-asset-design
Design and validate GitHub Copilot customization assets including agents, skills, instructions, and prompts. Provides architecture patterns, quality criteria, and integration strategies. Use when creating or improving Copilot customizations.
SKILL.md
| Name | copilot-asset-design |
| Description | Design and validate GitHub Copilot customization assets including agents, skills, instructions, and prompts. Provides architecture patterns, quality criteria, and integration strategies. Use when creating or improving Copilot customizations. |
CopilotCustomizer
Transform any codebase into an AI-enhanced development environment
GitHub Copilot is powerful, but generic. Your team needs AI that understands your stack, your patterns, and your standards. CopilotCustomizer generates a complete, production-ready customization layer in minutes—no configuration complexity, no manual setup.
Five powerful asset types working together: Cross-platform Skills + VS Code Agents + Coding Instructions + Slash Commands + Enterprise Standards
What You Get
- 🌍 Cross-Platform Skills - Portable AI capabilities that work in VS Code, GitHub CLI, Claude, Cursor, and more. Define domain knowledge once, use everywhere. Based on the open agentskills.io standard.
- 🤖 VS Code Agents - Specialized AI experts with orchestrated workflows. Automatic handoffs, parallel execution, and tool access tailored to your tech stack.
- 📋 Instructions - Project-specific coding standards that guide all AI interactions. Automatically applied to matching files—no manual prompting needed.
- ⚡ Prompts - Interactive slash commands for common workflows. Structured templates with variables, parameters, and refinement options.
- 🏢 Enterprise Standards - Organization-wide coding conventions that naturally influence generated assets. Define once, enforce everywhere.
Why CopilotCustomizer?
Problems it solves:
- ❌ Generic AI responses that don't match your codebase patterns or conventions
- ❌ Repeating the same context and requirements in every Copilot conversation
- ❌ Inconsistent code quality across team members and AI-generated code
- ❌ No systematic way to encode tribal knowledge into your AI development tools
- ❌ Time-consuming manual effort to maintain AI customizations as your stack evolves
What makes it different:
- ✅ Autonomous generation - Analyzes your repo and generates tailored assets automatically in 3-4 minutes
- ✅ True orchestration - Agents coordinate via handoffs with context preservation, not just isolated specialists
- ✅ Skills-first architecture - Portable capabilities work across VS Code, GitHub CLI, Claude, Cursor, and beyond
- ✅ Quality built-in - Automatic schema validation, cross-reference binding, and integrity verification
- ✅ Enterprise-ready - Standards system for organization-wide consistency and scalable governance
Perfect for:
- Teams wanting AI that deeply understands their codebase and coding standards
- Solo developers seeking consistent, high-quality AI assistance across projects
- Organizations needing to encode best practices into AI-powered development workflows
Quick Start
- Add to workspace:
File→Add Folder to Workspace→ Select CopilotCustomizer directory - Generate assets: Open Copilot Chat (
Ctrl+Shift+IorCmd+Shift+I) and run:/Bootstrap REPOSITORY_PATH: "/absolute/path/to/your-project" - Confirm generation: Review the analysis, then type
confirm(or click the confirm button in VS Code 1.109+)
The framework analyzes your repository, detects your tech stack, and generates Skills, Agents, Instructions, Prompts, and Standards tailored to your codebase—all saved in your-project/.github/.
Time: ~3-4 minutes | Result: Production-ready AI customization | No code changes: Only adds AI guidance files
Documentation
| Guide | Purpose |
|---|---|
| QUICKSTART.md | 5-minute setup guide ← Start here |
| EXAMPLES.md | Real-world examples with React, Python, .NET, Node.js, and more |
| HOW-TO.md | Complete commands reference + troubleshooting |
| docs/ARCHITECTURE.md | Architecture deep-dive + asset inventory |
| docs/CHANGELOG.md | Release history and breaking changes |
All commands work via slash commands in Copilot Chat or natural language:
# Complete setup - analyzes repo and generates all assets
/Bootstrap REPOSITORY_PATH: "/path/to/project"
# Analyze project - understand tech stack and recommend improvements
/Review TARGET_PATH: "/path/to/project"
# Create individual assets
/NewSkill SKILL_NAME: "name", PURPOSE: "description"
/NewAgent AGENT_NAME: "name", DOMAIN: "area"
/NewInstructions DOMAIN: "area", APPLY_TO: "**/*.ts"
/NewPrompt PURPOSE: "template goal"
# Advanced workflows
/NewMultiAgent SYSTEM_NAME: "name", SYSTEM_PATTERN: "linear"
/NewMultiAgent SYSTEM_NAME: "name", SYSTEM_PATTERN: "orchestra"
/NewAgentsFile REPOSITORY_PATH: "/path/to/project"
Tip: Type / in Copilot Chat to see all available commands with auto-completion.
See HOW-TO.md for detailed command documentation and advanced usage patterns.
Architecture
CopilotCustomizer uses an orchestrated agent system where a unified orchestrator (CopilotCustomizer agent) programmatically routes all work—including both external repository operations and toolkit self-improvement—to specialized subagents via the agent tool.
Core Components:
- Unified Orchestrator — Routes all workflows (user repos + toolkit) to specialized subagents
- 7 Agents — CopilotCustomizer (orchestrator), Bootstrap, Planner, Generator, Editor, Verifier, Evolve
- 6 Core Skills — repo-analysis, planning, asset-design, documentation, deployment-automation, orchestration
- 12 Instructions — Generation frameworks, quality patterns, standards resolution, toolkit maintenance
- 11 Prompts — Interactive commands for workflows, asset generation, maintenance, and toolkit evolution
- Enterprise Standards — Organization-wide conventions (43 standards across 8 categories)
Key Patterns:
- Multi-Workspace Pattern — CopilotCustomizer stays separate, generates into your projects
- Skills-First Strategy — Cross-platform analysis and planning before VS Code-specific operations
- Quality Gates — Automatic schema validation, harmonization, and integrity verification
- Programmatic Orchestration — All subagent invocation via
agenttool (no manual handoffs)
See docs/ARCHITECTURE.md for complete architecture documentation:
- Full asset inventory with file counts and purposes
- Enterprise standards system details
- Agent handoff workflows and orchestration patterns
- Bootstrap automation workflow breakdown
Maintenance & Toolkit Evolution
All maintenance and toolkit evolution commands are available as standard slash commands:
/QuickFix— Fast minimal-diff changes with single approval gate/Evolve— Monitor VS Code Copilot releases, audit toolkit assets, review documentation, optimize performance, implement feature requests, and maintain assets (optimize, harmonize, validate)
Toolkit Self-Improvement: Use /Evolve to evolve the framework itself. The CopilotCustomizer orchestrator routes these requests to the Evolve specialist subagent.
Requirements
Recommended:
- VS Code 1.109+ - Full feature support including Skills (GA), orchestration controls, parallel subagent execution, and enhanced handoffs
- GitHub Copilot subscription (Individual, Business, or Enterprise)
Minimum:
- VS Code 1.106+ with GitHub Copilot extension
No additional dependencies required - Framework uses pure Markdown with YAML frontmatter.
License
MIT License - See LICENSE for details.
Built for modern AI-powered development | Enterprise-ready | Open Source