terminy-procesowe-pl
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of instructional markdown and YAML configuration. No scripts (Python, JavaScript, etc.) are included, significantly reducing the attack surface for code execution or system compromise.
- [REMOTE_CODE_EXECUTION]: While the skill references an external connector
sejm-eli-mcp, this is described as a connector for querying official legal databases (ISAP/ELI). Under [TRUST-SCOPE-RULE], fetching documentation from official or well-known legal resources is treated as safe. There are no patterns of piping remote content to a shell. - [PRIVILEGE_ESCALATION]: The skill explicitly defines boundaries, stating it cannot write to calendars or send notifications. The
requires-human-approval: trueflag and the 'Fail-closed' philosophy ensure that the AI cannot take autonomous actions that would affect the user's system or legal standing. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user input (case dates, delivery methods). It implements boundary protection by requiring the agent to use a specific 'Term Card' template and strictly validating every step of the calculation against primary legal sources via the
sejm-eli-mcpskill, which mitigates the risk of data-driven instruction overrides. - [SAFE]: The skill follows security best practices: it uses local data residency, enforces human-in-the-loop verification for high-stakes tasks, and uses a standard open-source license (Apache-2.0).
Audit Metadata