How C-DAD Works

Overview

The cdad CLI is the part of Contract-Driven AI Development you actually run. Five commands cover the whole loop. cdad checkscores how legible your repo is right now and writes a report.cdad roadmap reads that report and ranks which capabilities are worth extracting first. cdad init scaffolds the contract files for whichever capability you picked. cdad validate keeps those files honest as the code underneath them changes, andcdad graph turns everything you've contracted so far into a dependency map you can actually look at.

None of these commands are meant to run once. The intended shape is a loop: check, roadmap, init, validate, graph, then check again, so you can watch the score move as the repo picks up more contracts over time. The pages below cover each command on its own, plus a walkthrough of writing your first contract end to end.

Pages in this area

• cdad check

  Scores repo agent-readiness from zero to ten and writes cdad-report.md, the file every later step reads.

• cdad roadmap

  Turns the check report into a ranked list of which capabilities to contract first.

• cdad init

  Scaffolds the contract triple, yaml, json, and markdown, for a single capability.

• cdad validate

  Checks contract schema, versioning, and generated-artifact consistency before a PR merges.

• cdad graph

  Renders every contracted capability and its dependencies as Mermaid, JSON, and Markdown.

• Writing your first contract

  A full walkthrough of the loop, from a cold repo to a validated, graphed contract.