secret-scan-gitleaks

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Command Execution] (MEDIUM): The skill executes shell commands (gitleaks detect) using parameters like --source=<path>. If the agent fails to sanitize the path input, it could lead to shell command injection.
  • [Indirect Prompt Injection] (HIGH): This is the primary risk. The skill ingests untrusted data from a codebase (file contents, commit messages, and metadata). Since the agent is instructed to 'Parse the results' and 'Summarize', an attacker could hide malicious prompt instructions within a scanned file or a fake secret. When the agent reads the resulting JSON report, it might obey instructions embedded in the 'secret' or 'author' fields.
  • [Data Exposure] (LOW): By design, this skill accesses sensitive data (hardcoded secrets). It includes a specific instruction to redact values (first 4 and last 2 characters), which is a positive safety control for the user interface, though the underlying agent context still handles the full JSON output.
  • [External Downloads] (LOW): The prerequisites suggest installing Gitleaks via brew or GitHub releases. While these are trusted sources for security professionals, the skill relies on the presence of external binary dependencies.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 12:59 PM