Article 11 requires a technical documentation file for high-risk systems before market placement. It must cover the system's intended purpose, architecture, data, risk-management measures and verification results. Writing it by hand is slow; generating it from your actual codebase is fast and far harder to challenge.
The dossier includes: a general description (intended purpose, input/output, deployers), a detailed architecture and development process description, training/validation/test data descriptions and governance, monitoring and logging design, accuracy and robustness evidence, risk-management measures, and instructions for use. Providers must keep it up to date and available to authorities.
Auditors cross-check the dossier against the shipped system. A documentation file that describes features your code does not implement is worse than no file at all — it looks like an attempt to mislead. Evidence generated from the actual codebase survives that check.
An automated pipeline reads the repo and drafts each dossier section directly from code, config and metadata: architecture from the module graph, model definitions from config, data handling from the data layer, and logging from the observability hooks. You review and sign, instead of writing from scratch.
See your score in 2 minutes. No signup, no code stored.
Scan Your Codebase NowYes — for the legal conformity assessment and declaration of conformity. The generator handles the technical evidence layer.
The scan scores each required documentation element so you know which sections are missing before an inspector does.
Recurring plans regenerate the dossier on every scan, so the file matches the latest deployed code.