Back to directory

SpecDD

SpecDD is a spec-driven development framework that keeps small, human-readable specification files beside the code they describe, giving both developers and AI coding agents one shared source of intent, boundaries and completion criteria.

SpecDD

SpecDD's specification files (with an .sdd extension) live next to the infrastructure, workflow or code they describe, and record intent, architecture, behavior, boundaries and local tasks in a format both a human and an AI agent can read directly. A command-line interface adds the framework's bootstrap files to a project, after which an agent is told to read the bootstrap file if it does not already do so automatically, and treats the specs as local source-of-truth context while implementing changes.

It supports both greenfield projects, where specs can be written before implementation starts, and existing codebases, where specs are introduced gradually around the areas actively being changed. It is described as agent-agnostic, working with any coding assistant that can read repository files, including Claude Code, Cursor, Copilot and several others, with no vendor-specific format or hosted memory.

It is written mostly in HTML and TypeScript tooling and released under Apache-2.0.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.