advisory-summary
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security threats were identified. The skill is composed entirely of natural language instructions and structural templates for generating legal memos.
- [NO_CODE]: The skill does not contain any executable scripts, shell commands, or external package dependencies. It operates solely through the agent's internal reasoning capabilities based on the provided workflow.
- [PROMPT_INJECTION]: Analysis of the instructions shows no evidence of jailbreak attempts, safety filter bypasses, or instructions to ignore previous system prompts. The language is professional and focused on legal drafting tasks.
- [DATA_EXFILTRATION]: There are no commands or network patterns (such as curl or wget) that would allow the exfiltration of data to external servers.
- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to process external data (legal texts and statutes), it lacks any 'write' or 'execute' capabilities that could be exploited if malicious instructions were embedded in that data. The output is strictly limited to a text-based template.
Audit Metadata