Skip to main content
Submit

GPT Engineer

Open-source AI that builds codebases from a spec — generate entire projects from one prompt

AgentsFree

GPT Engineer is an open-source AI coding agent that generates entire codebases from a natural language specification. Describe what you want to build, and GPT Engineer asks clarifying questions, then generates all files needed for a working project. The open-source CLI tool has 50K+ GitHub stars; gptengineer.app is the hosted web version with a browser-based editor.

Visit site
Key specs
54,964 GitHub stars source
as of 2026-11-12

Maker of GPT Engineer? Get your db.fyi badge

What we know

PricingFreefrom our original listing, not yet re-checked

Why it matters

  • 50K+ GitHub stars makes it one of the most-starred AI coding projects ever — community validation for AI-generated codebases.
  • Interactive clarification step before generation reduces ambiguity — the AI asks before assuming, producing better-fit output.
  • Generates entire project structures, not just functions — starter point for applications that would take days to scaffold manually.
  • Open source and local means code generation happens with your own API key — no data shared beyond what goes to OpenAI/Anthropic.

Key capabilities

  • Full project generation: Generate complete directory structure, all source files, config, and README from a spec.
  • Clarifying questions: AI asks targeted questions before generating to resolve ambiguities in the spec.
  • Step-by-step generation: Generates files in logical order with explanations for design decisions.
  • Iterative improvement: Continue providing feedback and the agent improves the generated codebase.
  • Model agnostic: Works with GPT, Claude, and other LLMs via API.
  • Local execution: Run the CLI with your own API key — code stays on your machine.
  • Web interface: gptengineer.app offers browser-based generation with visual file editor.
  • Language flexibility: Generates Python, JavaScript, TypeScript, Rust, Go, and any language the underlying LLM knows.

Technical notes

  • License: MIT (open source)
  • GitHub: github.com/gpt-engineer-org/gpt-engineer (50K+ stars)
  • CLI install: pip install gpt-engineer
  • Models: GPT and Claude (bring your own API key)
  • Web version: gptengineer.app (hosted; freemium)
  • OS: Linux, macOS, Windows

Ideal for

  • Developers who want to rapidly scaffold new projects — generate the structure, then customize and extend.
  • Hackathon teams who need a working starting point for a new application in hours.
  • Developers exploring AI-assisted project generation who want to self-host with their own API key.

Not ideal for

  • Generating production-ready, security-hardened code — AI-generated code needs thorough review.
  • Extending or modifying existing large codebases — GPT Engineer excels at new projects, less so at deep codebase editing.
  • Teams that need collaborative, cloud-hosted generation with team features — Bolt.new or v0 have better web interfaces.

See also

  • Devika — More capable autonomous AI agent for complex, multi-step tasks.
  • Bolt.new — Browser-based full-stack generation with live preview and deployment.
  • Cursor — AI IDE for controlled, iterative code generation and editing.

FAQ

What is GPT Engineer?

GPT Engineer is a project where you write a specification (a prompt describing the software you want), and the AI generates the entire codebase — all files, directories, and configurations needed for a working application. The open-source CLI tool (github.com/gpt-engineer-org/gpt-engineer) is the original project; gptengineer.app is a hosted web version with a visual editor.

Is GPT Engineer free?

The open-source CLI version is completely free (MIT license) — you provide your own OpenAI API key. The hosted web app (gptengineer.app) has a free tier with limited generations; paid plans unlock more usage and team features. The CLI + your own API key is the most economical for frequent use.

How does GPT Engineer work?

GPT Engineer follows a multi-step process: (1) You write a `prompt` file describing what to build; (2) GPT Engineer asks clarifying questions to reduce ambiguity; (3) It generates a task breakdown; (4) It writes all code files in sequence. The process is interactive — you can provide feedback and it iterates. The main strength is generating the initial project structure and boilerplate.

What is gptengineer.app vs. the CLI?

gptengineer.app has been rebranded to Lovable (lovable.dev) — a hosted web app with a visual editor where you can see generated code, make changes, and deploy. The CLI version (github.com/gpt-engineer-org/gpt-engineer) is a separate open-source tool for developers who want to run it locally. The CLI is more customizable; Lovable is more accessible for non-developers.

Alternatives

Integrations

None listed.

Built on

Good for these tasks

Related tools

Listed 2026-02-28 · updated 2026-11-12