Turn "lousy" AI outputs into production-grade code.
A CLI tool that scaffolds projects with the structure AI coding assistants need to be effective. Run npx lousy-agents init to create a new project with testing, linting, and GitHub Copilot configuration. Run npx lousy-agents copilot-setup in existing projects to generate a workflow that gives Copilot your environment context.
Lousy Agents is a CLI scaffolding tool that sets up your projects with the structure, instructions, and feedback loops that AI coding assistants need to be effective. One command gives you a production-ready development environment with testing, linting, and AI assistant configuration.
# Scaffold a new webapp project (no install required)
npx lousy-agents init --kind webapp
# Or use interactive mode to choose your project type
npx lousy-agents init
# Generate GitHub Copilot setup workflow from your project configuration
npx lousy-agents copilot-setup- Software Engineers: Frustrated by inconsistent AI output and looking for proven patterns to improve results.
- Curious Beginners: Interested in AI-assisted coding but unsure how to set things up for success.
- Team Leads: Exploring how to standardize AI tooling across a team or project.
- Platform Engineers: Need to automate project scaffolding in scripts or CI/CD pipelines.
No prior experience with coding agents is required—just curiosity and a willingness to experiment.
AI coding assistants work best when given clear constraints. Without structure, they guess—and often guess wrong. Lousy Agents provides the scaffolding they need to succeed:
- Instructions & Specs: Templates that clearly communicate your intent, so agents produce code that matches your vision.
- Feedback Loops: Pre-configured testing (Vitest) and linting (Biome) that let agents catch and fix their own mistakes immediately.
- Copilot Configuration: Settings and workflows that ground AI assistants in your specific engineering standards.
init- Scaffold new projects with testing, linting, and Copilot configurationcopilot-setup- Generate GitHub Actions workflows for Copilot environment setup
- MCP Server - Model Context Protocol server for AI assistant integration
A methodology where you write clear specifications first, giving agents precise requirements to implement—rather than vague prompts. Each scaffolded project includes instruction files for writing specs and tests.
Use the --kind flag to skip prompts and integrate into scripts or automation:
npx lousy-agents init --kind webapp # No prompts, perfect for CI/CDNo installation required! Use npx to run directly:
npx lousy-agents initFor frequent use, install globally:
npm install -g lousy-agentsFor detailed documentation on each command, see:
initcommand - Scaffold new projectscopilot-setupcommand - Generate Copilot workflows- MCP Server - AI assistant integration
Create a new webapp:
npx lousy-agents init --kind webappGenerate Copilot setup workflow:
npx lousy-agents copilot-setup| Feature | Status |
|---|---|
| Scaffolding for webapps | In Progress |
| Scaffolding for CLI | Not Started |
| Scaffolding for REST APIs | Not Started |
| Scaffolding for GraphQL APIs | Not Started |
| MCP server package | ✅ Complete |
initCommand - Project scaffoldingcopilot-setupCommand - Workflow generation- MCP Server - AI assistant integration
The ui/copilot-with-react directory contains a fully working reference implementation demonstrating these patterns in action. It's a Next.js + TypeScript project with:
- Pre-configured testing (Vitest) and linting (Biome)
- GitHub Copilot instructions and specs
- Dev Container configuration for GitHub Codespaces
Launch a GitHub Codespace to instantly spin up this environment and experiment with spec-driven development.
![@dependabot[bot]](https://avatars.githubusercontent.com/in/29110?s=64&v=4){kind=small}
