security-threat-model

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill defines a process for analyzing external project documentation such as feature briefs and specifications. While this creates a surface for indirect prompt injection, the instructions leverage this data solely for structured security analysis (Ingestion points: §1 Scope extraction). The skill includes no patterns attempting to bypass safety guardrails or override system instructions.
  • [COMMAND_EXECUTION]: The skill uses shell commands for legitimate administrative tasks, such as retrieving git repository metadata (e.g., git rev-parse) to generate unique paths for audit logs. This use of the Bash tool is strictly scoped to environmental inspection for reporting purposes.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes external fetching capabilities to retrieve security references from well-known and authoritative sources, including OWASP guidelines, NIST standards, and CVE databases. These operations are intended for security research and calibration of threat severity levels.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 08:13 AM
Security Audit — agent-trust-hub — security-threat-model