data-guide

Pass

Audited by Gen Agent Trust Hub on Apr 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill functions exclusively as a routing assistant and documentation guide. It does not include any scripts, shell commands, or network requests.
  • [PROMPT_INJECTION]: No patterns of instruction overriding, safety bypass, or role-play injection were detected. While the skill ingests user input regarding data goals, it lacks exploitable capabilities (Evidence: 1. Ingestion: User queries in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: None. 4. Sanitization: Absent).
  • [DATA_EXFILTRATION]: There is no access to sensitive system paths or credentials, and no network exfiltration patterns (curl, wget, etc.) are present.
  • [REMOTE_CODE_EXECUTION]: The skill does not define any external package dependencies or dynamic code evaluation logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 5, 2026, 08:06 PM
Security Audit — agent-trust-hub — data-guide