NYC

docs-init

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection (HIGH): The skill is designed to ingest and summarize codebase content which constitutes untrusted data. \n
  • Ingestion Points: Scout agents read files from the local filesystem based on ls -la results. \n
  • Boundary Markers: Absent; there are no instructions to the agents to ignore embedded commands or instructions within the source code being analyzed. \n
  • Capability Inventory: The skill can spawn sub-agents (Task tool) and write/update multiple files (docs/ directory and README.md). \n
  • Sanitization: Absent; external content is processed and used to influence the documentation manager without filtering.\n- Command Execution (LOW): Executes ls -la to identify the project structure, which is a necessary but low-risk administrative command in this context.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 01:01 PM