CANDOR.md

The standard introduces a dedicated AI‑DECLARATION.md file that lives alongside other repository documentation. It uses YAML front‑matter to capture structured metadata about how artificial‑intelligence tools have contributed to a project, followed by a required “## Notes” section for human‑readable context. At a minimum the file records a version identifier, a global usage level, and the notes; optional sections can detail individual processes (such as architecture, implementation, testing, documentation, review, and deployment) and specific components (files or directories) with their own levels.

Usage levels are defined by combinations of actions (act, prompt) and participants (Human, AI) to describe the degree of AI involvement, ranging from “none” (pure human work) to “auto” (AI completes the task autonomously). The specification also provides a formal YAML schema for validation, enabling tooling to automatically check compliance.

The goal is to make AI‑generated or AI‑assisted code transparent, allowing reviewers to pinpoint which parts of a codebase were produced with AI assistance and to verify them as needed. By standardizing this declaration, the project aims to foster a common convention for AI usage disclosure across software repositories.