secret-scan-trufflehog

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill processes data from external repositories which could contain malicious instructions designed to mislead the AI agent. Evidence: 1. Ingestion points: Git repositories and local filesystems. 2. Boundary markers: Findings are summarized from JSON output without explicit delimiters to isolate untrusted data. 3. Capability inventory: Execution of the trufflehog command-line tool. 4. Sanitization: Includes a requirement to redact raw secrets before displaying them.
  • [Command Execution] (SAFE): The skill uses templates for running shell commands. While this introduces a surface for command injection if inputs are not handled carefully by the agent, the behavior is necessary for the skill's primary purpose of security auditing.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:46 PM