security-auditor

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple security CLI tools and shell commands for system reconnaissance and vulnerability scanning. It also runs a specific local Python module for LLM triage.
  • [EXTERNAL_DOWNLOADS]: Fetches updated security rule packs from the Semgrep registry during the scanning phase.
  • [PROMPT_INJECTION]: Vulnerable to indirect prompt injection because untrusted code snippets from the scanned repository are provided directly to an LLM for evaluation.
  • Ingestion points: Local source files and configuration files discovered during Phases 1-3.
  • Boundary markers: Triple backticks are used in the triage prompt to delimit code context.
  • Capability inventory: Extensive shell execution (SAST tools), filesystem writes for reporting, and network communication for LLM triage.
  • Sanitization: No explicit sanitization of code snippets before model processing.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 03:58 AM
Security Audit — agent-trust-hub — security-auditor