@tank/idd

Intent-driven development (IDD) prioritizes defining human requirements and architectural constraints in an INTENT.md file before code generation, ensuring AI-assisted software creation aligns precisely with the desired outcome and reducing development waste.

You use the IDD lifecycle to define API contracts and behavioral examples in YAML frontmatter and Markdown, enabling structured interviewing, design critique, and phased TDD execution to scaffold a new microservice before generating code.

IDD is ideal for greenfield software creation where intent can be defined upfront. For existing codebases, implementing strict intent file structures and architectural constraints requires defining current behaviors in INTENT.md before applying automated drift detection.

Code-intent drift detection is an automated mechanism that synchronizes generated code with the original INTENT.md specification. You need it during AI-assisted development to enforce strict rules and ensure the output does not diverge from the defined architectural constraints.

Yes, the IDD framework utilizes YAML frontmatter and Markdown within INTENT.md to specify strict rules on file structure and content, ensuring the AI respects architectural constraints and API contracts during the code generation lifecycle.

The limitation of intent-driven development is its strict reliance on upfront specification; if the INTENT.md lacks meticulous detail or behavioral examples, the AI-generated code may fail to meet complex architectural constraints, requiring continuous structured interviewing and locking mechanisms to maintain alignment.