technology-stack-blueprint-generator — GitHub Copilot skill
This skill scans a codebase and builds a blueprint of its stack, architecture, and coding practices. Use it to document an existing project before making changes or generating more code.
npx skills add github/awesome-copilot --skill technology-stack-blueprint-generatorWhat technology-stack-blueprint-generator does
The skill looks through project files, configuration, dependencies, build scripts, and pipelines to identify languages and tooling. Its configurable blueprint can use Markdown, JSON, YAML, or HTML. Choose an analysis depth and decide whether to include versions, licenses, diagrams, usage examples, and coding conventions.
Separate analysis sections cover .NET, Java, JavaScript, React, and Python. Depending on the options and what the codebase contains, the result can describe project structure, libraries, testing approaches, error handling, and data access patterns. The output reflects the files it can inspect.
Install it when you need a written map of an existing codebase or a reference for consistent implementation. For a small project with a sufficient README, this level of documentation may be more than you need.
What it helps with
- identify project languages and dependencies
- document the codebase structure and architecture
- extract examples of established implementation patterns
- generate a stack blueprint in Markdown, JSON, YAML, or HTML
Example prompts for the agent
- Analyze this repository and create a technology stack blueprint in Markdown.
- Document the architecture in detail, including versions and diagrams.
- Show the project's coding conventions and API and data access examples.
How to install technology-stack-blueprint-generator
1Skills CLI (Vercel)
npx skills add github/awesome-copilot --skill technology-stack-blueprint-generator2GitHub CLI
gh skill install github/awesome-copilot technology-stack-blueprint-generator3Manually: put the files in place
.github/skills/technology-stack-blueprint-generator/SKILL.md1Skills CLI (Vercel)
npx skills add github/awesome-copilot --skill technology-stack-blueprint-generator -a claude-code2Manually: put the files in place
~/.claude/skills/technology-stack-blueprint-generator/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/technology-stack-blueprint-generator/
Manually: put the files in place
.agents/skills/technology-stack-blueprint-generator/SKILL.mdFor all projects use ~/.agents/skills/
Manually: put the files in place
.cursor/skills/technology-stack-blueprint-generator/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
Platform CLI
gemini skills install https://github.com/github/awesome-copilot --path skills/technology-stack-blueprint-generatorManually: put the files in place
.devin/skills/technology-stack-blueprint-generator/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
From the same set554
The whole set →Similar
Questions about technology-stack-blueprint-generator
How do I install technology-stack-blueprint-generator in GitHub Copilot?
Run npx skills add github/awesome-copilot --skill technology-stack-blueprint-generator. You can also use gh skill install github/awesome-copilot technology-stack-blueprint-generator.
What does technology-stack-blueprint-generator do?
It analyzes a codebase and creates a blueprint describing its technologies, dependencies, and development practices. You can configure the output format and analysis depth.
Which technologies does the skill analyze?
It has dedicated analysis sections for .NET, Java, JavaScript, React, and Python. You can also select Auto-detect to identify the stack from project files.
Source text
SKILL.md · MIT
---
name: technology-stack-blueprint-generator
description: 'Comprehensive technology stack blueprint generator that analyzes codebases to create detailed architectural documentation. Automatically detects technology stacks, programming languages, and implementation patterns across multiple platforms (.NET, Java, JavaScript, React, Python). Generates configurable blueprints with version information, licensing details, usage patterns, coding conventions, and visual diagrams. Provides implementation-ready templates and maintains architectural consistency for guided development.'
---
# Comprehensive Technology Stack Blueprint Generator
## Configuration Variables
${PROJECT_TYPE="Auto-detect|.NET|Java|JavaScript|React.js|React Native|Angular|Python|Other"} <!-- Primary technology -->
${DEPTH_LEVEL="Basic|Standard|Comprehensive|Implementation-Ready"} <!-- Analysis depth -->
${INCLUDE_VERSIONS=true|false} <!-- Include version information -->
${INCLUDE_LICENSES=true|false} <!-- Include license information -->
${INCLUDE_DIAGRAMS=true|false} <!-- Generate architecture diagrams -->
${INCLUDE_USAGE_PATTERNS=true|false} <!-- Include code usage patterns -->
${INCLUDE_CONVENTIONS=true|false} <!-- Document coding conventions -->
${OUTPUT_FORMAT="Markdown|JSON|YAML|HTML"} <!-- Select output format -->
${CATEGORIZATION="Technology Type|Layer|Purpose"} <!-- Organization method -->
## Generated Prompt
"Analyze the codebase and generate a ${DEPTH_LEVEL} technology stack blueprint that thoroughly documents technologies and implementation patterns to facilitate consistent code generation. Use the following approach:
### 1. Technology Identification Phase
- ${PROJECT_TYPE == "Auto-detect" ? "Scan the codebase for project files, configuration files, and dependencies to determine all technology stacks in use" : "Focus on ${PROJECT_TYPE} technologies"}
- Identify all programming languages by examining file extensions and content
- Analyze configuration files (package.json, .csproj, pom.xml, etc.) to extract dependencies
- Examine build scripts and pipeline definitions for tooling information
- ${INCLUDE_VERSIONS ? "Extract precise version information from package files and configuration" : "Skip version details"}
- ${INCLUDE_LICENSES ? "Document license information for all dependencies" : ""}
### 2. Core Technologies Analysis
${PROJECT_TYPE == ".NET" || PROJECT_TYPE == "Auto-detect" ? "#### .NET Stack Analysis (if detected)
- Target frameworks and language versions (detect from project files)
- All NuGet package references with versions and purpose comments
- Project structure and organization patterns
- Configuration approach (appsettings.json, IOptions, etc.)
- Authentication mechanisms (Identity, JWT, etc.)
- API design patterns (REST, GraphQL, minimal APIs, etc.)
- Data access approaches (EF Core, Dapper, etc.)
- Dependency injection patterns
- Middleware pipeline components" : ""}
${PROJECT_TYPE == "Java" || PROJECT_TYPE == "Auto-detect" ? "#### Java Stack Analysis (if detected)
- JDK version and core frameworks
- All Maven/Gradle dependencies with versions and purpose
- Package structure organization
- Spring Boot usage and configurations
- Annotation patterns
- Dependency injection approach
- Data access technologies (JPA, JDBC, etc.)
- API design (Spring MVC, JAX-RS, etc.)" : ""}
${PROJECT_TYPE == "JavaScript" || PROJECT_TYPE == "Auto-detect" ? "#### JavaScript Stack Analysis (if detected)
- ECMAScript version and transpiler settings
- All npm dependencies categorized by purpose
- Module system (ESM, CommonJS)
- Build tooling (webpack, Vite, etc.) with configuration
- TypeScript usage and configuration
- Testing frameworks and patterns" : ""}
${PROJECT_TYPE == "React.js" || PROJECT_TYPE == "Auto-detect" ? "#### React Analysis (if detected)
- React version and key patterns (hooks vs class components)
- State management approach (Context, Redux, Zustand, etc.)
- Component library usage (Material-UI, Chakra, etc.)
- Routing implementation
- Form handling strategies
- API integration patterns
- Testing approach for components" : ""}
${PROJECT_TYPE == "Python" || PROJECT_TYPE == "Auto-detect" ? "#### Python Analysis (if detected)
- Python version and key language features used
- Package dependencies and virtual environment setup
- Web framework details (Django, Flask, FastAPI)
- ORM usage patterns
- Project structure organization
- API design patterns" : ""}
### 3. Implementation Patterns & Conventions
${INCLUDE_CONVENTIONS ?
"Document coding conventions and patterns for each technology area:
#### Naming Conventions
- Class/type naming patterns
- Method/function naming patterns
- Variable naming conventions
- File naming and organization conventions
- Interface/abstract class patterns
#### Code Organization
- File structure and organization
- Folder hierarchy patterns
- Component/module boundaries
- Code separation and responsibility patterns
#### Common Patterns
- Error handling approaches
- Logging patterns
- Configuration access
- Authentication/authorization implementation
- Validation strategies
- Testing patterns" : ""}
### 4. Usage Examples
${INCLUDE_USAGE_PATTERNS ?
"Extract representative code examples showing standard implementation patterns:
#### API Implementation Examples
- Standard controller/endpoint implementation
- Request DTO pattern
- Response formatting
- Validation approach
- Error handling
#### Data Access Examples
- Repository pattern implementation
- Entity/model definitions
- Query patterns
- Transaction handling
#### Service Layer Examples
- Service class implementation
- Business logic organization
- Cross-cutting concerns integration
- Dependency injection usage
#### UI Component Examples (if applicable)
- Component structure
- State management pattern
- Event handling
- API integration pattern" : ""}
### 5. Technology Stack Map
${DEPTH_LEVEL == "Comprehensive" || DEPTH_LEVEL == "Implementation-Ready" ?
"Create a comprehensive technology map including:
#### Core Framework Usage
- Primary frameworks and their specific usage in the project
- Framework-specific configurations and customizations
- Extension points and customizations
#### Integration Points
- How different technology components integrate
- Authentication flow between components
- Data flow between frontend and backend
- Third-party service integration patterns
#### Development Tooling
- IDE settings and conventions
- Code analysis tools
- Linters and formatters with configuration
- Build and deployment pipeline
- Testing frameworks and approaches
#### Infrastructure
- Deployment environment details
- Container technologies
- Cloud services utilized
- Monitoring and logging infrastructure" : ""}
### 6. Technology-Specific Implementation Details
${PROJECT_TYPE == ".NET" || PROJECT_TYPE == "Auto-detect" ?
"#### .NET Implementation Details (if detected)
- **Dependency Injection Pattern**:
- Service registration approach (Scoped/Singleton/Transient patterns)
- Configuration binding patterns
- **Controller Patterns**:
- Base controller usage
- Action result types and patterns
- Route attribute conventions
- Filter usage (authorization, validation, etc.)
- **Data Access Patterns**:
- ORM configuration and usage
- Entity configuration approach
- Relationship definitions
- Query patterns and optimization approaches
- **API Design Patterns** (if used):
- Endpoint organization
- Parameter binding approaches
- Response type handling
- **Language Features Used**:
- Detect specific language features from code
- Identify common patterns and idioms
- Note any specific version-dependent features" : ""}
${PROJECT_TYPE == "React.js" || PROJECT_TYPE == "Auto-detect" ?
"#### React Implementation Details (if detected)
- **Component Structure**:
- Function vs class components
- Props interface definitions
- Component composition patterns
- **Hook Usage Patterns**:
- Custom hook implementation style
- useState patterns
- useEffect cleanup approaches
- Context usage patterns
- **State Management**:
- Local vs global state decisions
- State management library patterns
- Store configuration
- Selector patterns
- **Styling Approach**:
- CSS methodology (CSS modules, styled-components, etc.)
- Theme implementation
- Responsive design patterns" : ""}
### 7. Blueprint for New Code Implementation
${DEPTH_LEVEL == "Implementation-Ready" ?
"Based on the analysis, provide a detailed blueprint for implementing new features:
- **File/Class Templates**: Standard structure for common component types
- **Code Snippets**: Ready-to-use code patterns for common operations
- **Implementation Checklist**: Standard steps for implementing features end-to-end
- **Integration Points**: How to connect new code with existing systems
- **Testing Requirements**: Standard test patterns for different component types
- **Documentation Requirements**: Standard doc patterns for new features" : ""}
${INCLUDE_DIAGRAMS ?
"### 8. Technology Relationship Diagrams
- **Stack Diagram**: Visual representation of the complete technology stack
- **Dependency Flow**: How different technologies interact
- **Component Relationships**: How major components depend on each other
- **Data Flow**: How data flows through the technology stack" : ""}
### ${INCLUDE_DIAGRAMS ? "9" : "8"}. Technology Decision Context
- Document apparent reasons for technology choices
- Note any legacy or deprecated technologies marked for replacement
- Identify technology constraints and boundaries
- Document technology upgrade paths and compatibility considerations
Format the output as ${OUTPUT_FORMAT} and categorize technologies by ${CATEGORIZATION}.
Save the output as 'Technology_Stack_Blueprint.${OUTPUT_FORMAT == "Markdown" ? "md" : OUTPUT_FORMAT.toLowerCase()}'
"
© github · MIT · github/awesome-copilot