pdd-cli

What it is and what it does

PDD is a command-line system that inverts the traditional code-first development model: you write prompts (in `.prompt` files) as your primary source of truth, and PDD automatically generates, tests, and maintains code in your target language. The core workflow is the `sync` command, which orchestrates code generation from prompts, dependency injection, test creation, error fixing, and verification in a single batch operation. It supports Python, TypeScript, Go, Java, C++, and other languages as compilation targets.

The system is designed to reduce maintenance costs by treating prompts—not code—as the durable specification. When requirements change, you update the prompt and regenerate clean code rather than patching existing files. It runs in cloud mode (GitHub SSO, no local API keys) by default, or locally with your own LLM credentials. The 49 runtime dependencies include langchain, firebase, cryptography, and other libraries for LLM integration, web scraping, and async operations.

Use it for:

• Generate a complete application (e.g., HTML canvas app) from a single natural-language prompt specification.
• Automatically create and improve unit tests by analyzing code coverage and regenerating test suites from prompts.
• Maintain consistency across a large codebase by treating prompts as a single source of truth and regenerating all code when requirements shift.
• Reduce LLM costs by using batch-oriented workflows instead of interactive AI assistants, with optional batch API discounts.
• Collaborate across technical and non-technical teams by writing specifications in natural language rather than code.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

PDD is a CLI tool that treats natural-language prompts as source code, automatically generating, testing, and maintaining code in Python, TypeScript, Go, and other languages from `.prompt` files.

Yes, with conditions. PDD is actively maintained (last release 19 days ago), has low install friction, and offers a genuinely novel development model for teams willing to adopt prompt-driven workflows. The 49 dependencies are substantial but all standard. Install if you want to experiment with AI-native development, treat prompts as source, and regenerate code rather than patch it. Skip if you need a lightweight tool, prefer traditional code-first development, or require production stability guarantees—the project is in Alpha (Development Status 3) and cloud mode is on a waitlist.

Install

pdd-cli on PyPI

pip install pdd-cli

uv add pdd-cli

poetry add pdd-cli

Installing pdd-cli

Before you install

Low friction: pure Python wheel with no compiled dependencies. Active maintenance—last release 19 days ago, 840 repository stars. Requires Python 3.12+. The 49 runtime dependencies are substantial (langchain, firebase, firecrawl, cryptography, pandas) but all are standard PyPI packages; installation is straightforward.

License in practice

MIT license (permissive). You can use, modify, and distribute PDD freely in commercial and private projects with minimal restrictions—only attribution and license notice required.

Quickstart

pip install pdd-cli
pdd --version
pdd sync factorial_calculator

Requires Python 3.12 or later. Cloud mode requires GitHub SSO or internet access; local mode requires OPENAI_API_KEY, ANTHROPIC_API_KEY, or other LLM provider credentials in environment.

Verify before relying

• Whether the 49 runtime dependencies (including langchain, firebase_admin, firecrawl-py) are all required for basic sync/generate operations or if many are optional for specific features.
• Performance and token cost in practice for typical projects—fact sheet mentions batch discounts but no concrete benchmarks.
• Stability of the cloud mode (currently described as 'waitlist') and whether local-only workflows are production-ready.

Package facts

Evidence: pdd_cli-0.0.309-py3-none-any.whl

Keywords: prompt-driven development, code generation, AI, LLM, unit testing, software development

Tags

More Artificial Intelligence packages