ki-folgenabschaetzung-ki-governance-mandat

What it does

Breaks down the assessment task for KI-Folgenabschätzung in governance contexts, indicating how to determine outputs, deadlines, responsibilities, and evidentiary needs. Produces a Beweislast- und Substantiierungsmatrix (evidence burden and substantiation matrix) and an output document structured around governance, legal frame, and risk.

How it works

Follows a multistep workflow: clarify role and outputs, mark deadlines and risk, verify governing norms, identify responsible entities, gather documents and evidence, classify risk tracks (Vereinfacht vs Vollständig), identify legal acts and obligations, draft a comprehensive Abschätzung in a house format, compare against CLAUDE.md governance commitments, and finalize with deliverables (Abschätzungsdokument, Bedingungsliste, Übergabe-Flags).

When to use it

Triggered when performing KI-Folgenabschätzung in digital tool governance; applies across steps 0–4 and Mandate-Kontext to determine whether a mandate requires a full or simplified assessment and to populate outputs and governance checks.

What it can touch

Configures from a local CLAUDE.md setup and uses references to governance artifacts (FRIA, DSFA, Datenschutz etc.), and references to listed regulatory sections. It interacts with documents like FRIA, DSFA, Audit-Bericht, DSGVO-DPIA, Schulungsnachweis, and CLAUDE.md. It mentions tool: claude-code as a declared tool.

Caveats

Defines regulatory citations and checks but advises against using model-knowledge citations. Requires verification against primary sources; some fields rely on live policy references and may require human confirmation before deployment. No guarantees of outcome are stated; it outlines what to collect and how to structure the assessment.