Local Executable Coding Agent
Transform natural language prompts into complete, runnable software projects with automated planning, implementation, testing, and documentation.
Overview
This skill enables you to generate fully-functional software projects from simple descriptions. The agent handles the entire development lifecycle: project planning, file creation, dependency management, testing, and documentation. It supports multiple AI providers (OpenAI, vLLM) and generates production-ready code with proper structure and best practices.
Key Capabilities
**Complete Project Generation**: End-to-end project creation from natural language**Multi-Language Support**: Python, TypeScript/React, FastAPI, Express.js, and more**Automated Testing**: Test-fix iteration loops with comprehensive test suites**Modern Tooling**: UV for Python dependencies, proper package management**Security**: Command validation, path restrictions, sandboxed execution**Rich Output**: Interactive CLI with progress tracking and detailed logsInstructions
When the user requests a software project, follow this workflow:
1. Parse the Request
Extract key information from the user's prompt:
Project type (web app, API, CLI tool, etc.)Technology stack preferencesCore features and requirementsAny specific constraints or requirements2. Project Planning Phase
Create a detailed project plan including:
Project structure (directories and key files)Technology stack and dependenciesCore components and their responsibilitiesImplementation steps in logical orderTesting strategyPresent the plan to the user for confirmation before proceeding.
3. Implementation Phase
Execute the plan using these tools systematically:
**File Operations:**
Create project directories with proper structureWrite source files with complete implementationsGenerate configuration files (package.json, pyproject.toml, etc.)Create README.md with usage instructions**Dependency Management:**
For Python: Use UV for modern dependency managementFor Node.js: Generate proper package.json with all dependenciesInclude dev dependencies for testing and formatting**Code Quality:**
Follow language-specific best practicesInclude proper error handlingAdd type annotations (Python, TypeScript)Format code consistently4. Testing Phase
Generate and run comprehensive tests:
Unit tests for core functionalityIntegration tests for APIsEdge case coverageError handling verificationIf tests fail, analyze errors and fix issues iteratively.
5. Documentation Phase
Generate complete documentation:
README.md with setup instructions, usage examples, and API documentationInline code comments for complex logicConfiguration file documentationBuild and deployment instructions6. Project Delivery
Package the complete project:
Verify all files are createdConfirm tests passValidate documentation completenessProvide setup and run instructionsSupported Project Types
FastAPI Backend
SQLite database with SQLAlchemy ORMFull CRUD operationsPydantic models for validationComprehensive API documentationPytest test suiteReact TypeScript App
Modern React with hooksTypeScript for type safetyComponent-based architectureCSS modules or styled-componentsJest and React Testing LibraryPython CLI Tool
Click or argparse for CLIProper command structureError handling and validationComprehensive test coverageHelp documentationExpress.js API
RESTful API designAuthentication middlewareDatabase integrationAPI documentationJest test suiteExample Usage
**User Request:**
"Build a FastAPI todo list app with SQLite database"
**Agent Response:**
1. Creates project plan with database schema, API endpoints, and test strategy
2. Generates project structure with proper directories
3. Implements FastAPI application with CRUD endpoints
4. Creates SQLAlchemy models and database setup
5. Writes comprehensive pytest test suite
6. Generates README with setup and usage instructions
7. Validates all tests pass and documentation is complete
Important Constraints
**Security**: Never execute untrusted commands. Validate all file paths.**Dependencies**: Always specify exact versions for reproducibility**Testing**: All generated code must include tests**Documentation**: Every project must have clear setup instructions**Error Handling**: Implement proper error handling in all code**Path Safety**: Restrict file operations to project directoryConfiguration
The agent supports configuration for:
AI provider selection (OpenAI, vLLM)Model selection and parametersOutput directory locationSecurity sandbox settingsLogging verbosityTool Availability
The agent has access to 10+ tools including:
File creation and editingDirectory managementCommand execution (sandboxed)Code formatting and lintingTest executionPackage installationDocumentation generationQuality Standards
All generated projects must:
Follow language-specific style guidesInclude comprehensive error handlingHave clear, consistent naming conventionsBe properly typed (where applicable)Pass all generated testsInclude complete documentationUse modern, maintainable patterns