cra-meldepflichten
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources, such as Proof of Concept (PoC) code and vulnerability logs.\n
- Ingestion points: The 'Eingaben' section of
SKILL.mdidentifies input sources including 'Nachweise für die aktive Ausnutzung' (Proof of Concept, logs, and CERT reports).\n - Boundary markers: The skill does not implement specific delimiters or instructions to isolate the untrusted input from its core processing logic.\n
- Capability inventory: The skill is limited to generating text-based compliance reports and calculating deadlines; it does not possess capabilities for file writing, network access, or command execution.\n
- Sanitization: No explicit sanitization or input validation routines are specified for the external data ingested.\n- [EXTERNAL_DOWNLOADS]: The skill contains references to official government and regulatory domains for compliance documentation.\n
- Official Resources: It provides links to
eur-lex.europa.eufor EU legislative texts,digital-strategy.ec.europa.eufor Commission policies, andbsi.bund.defor German cybersecurity authority information.
Audit Metadata