avast-premium-security-malware-detection
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, including GitHub repository descriptions, metadata, and file lists. A malicious actor could craft repository descriptions or metadata containing instructions designed to confuse the agent or bypass the detection logic, potentially leading to incorrect security assessments.
- Ingestion points: Analyzes repository descriptions (
repo.description) and file system paths viarecursive_directory_iterator. - Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious commands embedded within the analyzed content.
- Capability inventory: The skill involves file system scanning, logical classification of threats, and generating text-based reports for external submission.
- Sanitization: The skill uses basic string normalization (lowercase conversion) for keyword matching but lacks sanitization or validation to prevent adversarial instructions from influencing agent behavior.
- [SAFE]: The skill references official download URLs for well-known security services (Avast, AVG, Norton, Kaspersky). These references are legitimate and do not involve data exfiltration or malicious redirects.
Audit Metadata