Skip to content

FAQ.md

Latest commit

 

History

History
124 lines (79 loc) · 3.19 KB

FAQ.md

File metadata and controls

124 lines (79 loc) · 3.19 KB

ODD Frequently Asked Questions

General

What is ODD?

ODD (Output-Driven Development) is a methodology for AI-native software engineering. It prioritizes Artifacts and Decision Responsibility over Code.

How is ODD different from GitHub Copilot?

Copilot ODD
AI assistant that helps you write code Framework that defines contracts for AI to generate code
You still review all code System verifies code automatically
Responsibility is vague Responsibility is clear via contracts
Focus on "how to write" Focus on "what to produce"

Is ODD a tool or a methodology?

ODD is a methodology with a reference implementation. You can practice ODD with any tools (Copilot, Cursor, Claude, etc.). This repo is a reference implementation.

Usage

What LLMs are supported?

Any OpenAI-compatible API:

  • OpenAI (GPT-4, GPT-4o)
  • NVIDIA (Llama, Mistral)
  • Anthropic (via adapter)
  • Local models (Ollama, LM Studio)

Do I need to know how to code?

No. ODD separates definition from execution:

  • Definition (contracts) requires domain knowledge, not coding
  • Execution (code generation) is done by AI

Domain experts can participate directly without programming skills.

Can I use ODD for [specific use case]?

ODD is designed for:

  • ✅ Enterprise applications
  • ✅ Finance/Healthcare (compliance requirements)
  • ✅ Large teams (standardization needs)
  • ✅ High-risk systems (auditability needs)

ODD is NOT designed for:

  • ❌ Creative exploration
  • ❌ Pure art projects
  • ❌ Zero-cost prototyping

Technical

What is a "contract"?

A contract is a precise agreement defining artifacts:

  • Inputs: What the artifact receives
  • Outputs: What the artifact produces
  • Implicit requirements: Expert knowledge encoded as rules
  • Verification hints: How to validate the result

What is "sealing"?

Sealing creates an immutable audit trail:

  • Calculates SHA-256 hash of each artifact
  • Establishes a complete hash chain
  • Enables traceability from final code back to requirements

What if verification fails?

Two scenarios:

  1. Critical failure (e.g., missing password hashing) → Block deployment, require human decision
  2. Warning (e.g., missing exception handling) → Log warning, allow override

This is Human-in-the-Loop Governance: AI generates, system screens, humans decide.

Contributing

How can I contribute?

  1. Add contract templates to artifacts/standards/
  2. Improve verification rules for existing templates
  3. Share your use cases in Discussions
  4. Report bugs via Issues

See CONTRIBUTING.md for details.

License

Can I use ODD commercially?

Yes. ODD is licensed under MIT License, which permits commercial use.

Do I need to credit ODD?

No, but it's appreciated if you do.

Contact

Author: Yi Fu (ODDFounder) Email: fuyi.it@live.cn WeChat: Fuyi-ODDFounder GitHub: @oddfounder

⭐ Found this helpful? Star us on GitHub

ODD: Unleash AI Speed. Reduce Engineering Risk.