I built idea-2-repo, a CLI that takes plain-English product ideas and turns them into a structured starter repository — complete with docs, TODOs, architecture outlines, and source scaffolding — by leveraging GitHub Copilot CLI’s natural-language coding agent directly in the terminal.
What I built
idea2repo aims to remove blank-project friction by generating a ready-to-use project skeleton from one simple idea prompt.
Core capabilities
✔ Idea normalization + project classification
✔ Architecture suggestion with Copilot CLI
✔ Scaffold generation (docs, TODOs, source files)
✔ Offline fallback (REASONING_BACKEND=offline)
✔ Dry-run mode for safe previews
1) Environment & Repo Context
I ensured tool versions and repository context were set up before generation.
2) Validation Checks
I verified the core validation pipeline end-to-end.
Lint & Build
Unit / Integration Tests
E2E Tests
3) Copilot CLI Usage
I confirmed Copilot CLI availability and used it to shape architecture ideas.
Copilot Version
Copilot Prompt Output
Prompt used:
copilot -p "Design a TypeScript CLI scaffold architecture in 5 concise bullets."
4) Real Project Generation
I ran a real project generation (non-dry run):
node dist/bin/idea2repo.js generate "AI-powered expense tracker for freelancers"
Generated artifacts verified:
✔ Project folder exists
✔ Docs and scaffold exist
✔ TODOs and architecture files present
5) Offline Fallback Demonstration
Verified offline mode behavior with:
REASONING_BACKEND=offline node dist/bin/idea2repo.js generate "Simple todo api for students" --dry-run
6) CLI UX / Help Output
CLI usability demonstration:
node dist/bin/idea2repo.js --help
What I Learned
Building idea2repo taught me a few practical lessons:
Copilot CLI shines at rapid architectural ideation when prompts are clear and scoped.
Offline fallback matters for reproducibility and deterministic runs.
Dry-run plus real-run improves confidence during demos and validation.
Repo
GitHub: https://github.com/GeoAziz/idea-2-repo
Top comments (0)