ai-act-triage-pl

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a purely instructional guide for regulatory compliance. It provides a multi-step workflow for classifying systems based on the EU AI Act without requesting sensitive permissions or performing dangerous actions.
  • [COMMAND_EXECUTION]: The skill does not contain any shell commands, scripts, or system-level execution patterns. It relies on standard information gathering and processing.
  • [DATA_EXFILTRATION]: No exfiltration patterns or hardcoded credentials were found. The skill configuration specifies local data residency and no PII egress, which aligns with the provided instructions.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external information (descriptions of systems to be triaged) and legal texts from EUR-Lex. While this creates an ingestion surface for untrusted data, the skill includes robust mitigation strategies, such as mandatory human approval ('requires-human-approval: true') and explicit instructions to verify all legal references against official sources ('Krok 0').
  • [PROMPT_INJECTION]: The instructions do not attempt to bypass safety filters or override agent behavior. The language is consistent with legal triage and compliance auditing.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 07:31 PM
Security Audit — agent-trust-hub — ai-act-triage-pl