security-threat-awareness

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading ClamAV source code from Cisco Talos's official GitHub repository. This is a reference to a well-known security project for educational purposes.
  • [COMMAND_EXECUTION]: Includes shell commands for installing and running the ClamAV antivirus tool using sudo. These are standard operational procedures for system-level security software and are provided in a defensive context.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides logic to evaluate repository metadata, which represents a potential surface for indirect prompt injection if untrusted data is processed.
  • Ingestion points: Repository metadata (e.g., description, owner name) processed by the is_repository_suspicious function in SKILL.md.
  • Boundary markers: Absent.
  • Capability inventory: Shell commands for freshclam and clamscan are documented in SKILL.md.
  • Sanitization: The provided Python logic includes heuristic checks and keyword filtering to validate repository data.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 04:52 AM
Security Audit — agent-trust-hub — security-threat-awareness