The manifesto
Four values, ten principles, the position in full. Signed by pull request. Read and sign it.
Product Definition as Code
Your product, defined like code. An open methodology for the upstream layer of the AI-SDLC.
The bottleneck moved left
Section titled “The bottleneck moved left”AI-assisted engineering has made implementation fast. The limiting factor of software is no longer writing code; it is knowing, precisely and verifiably, what the product is. An agent that writes code in minutes amplifies whatever understanding it is given, including none.
PDaC models product knowledge as a versioned, validated graph of small, related Markdown artifacts: actors, journeys, use cases, business rules, domain language and requirements. Humans and AI agents consume the same canonical model, and every implementation increment traces back to the product knowledge it serves.
While Spec-Driven Development tools define how a single change gets built, PDaC defines what the product is: the graph that outlives every spec.
The specification
Ten normative chapters, RFC 2119 language, stable diagnostic codes. Currently v0.9, request for comments. Start with the overview.
The reference implementation
ProductShape is to PDaC what OpenSpec is to Spec-Driven Development. Try it:
npx @prodshape/cli init.
ProductShape on GitHub.
Build an implementation
The spec is implementation-independent by design and wants more than one. Conformance and listing criteria.
Where it comes from
Section titled “Where it comes from”PDaC was introduced in Product Definition as Code for the AI-SDLC (July 2026) and extracted into an open specification. Founding supporter: Plain Concepts.