NYC

project-scaffold

Fail

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill processes untrusted user input (project and organization names) through interactive prompts in scaffold.py. Given the skill's high-tier capabilities, this represents a significant injection surface. 1. Ingestion points: interactive prompts in scaffold.py; 2. Boundary markers: Absent from documentation; 3. Capability inventory: Extensive file system modification and execution of shell commands (e.g., bun install); 4. Sanitization: Not verifiable.
  • [Command Execution] (HIGH): The skill instructs the agent to execute a local Python script (~/.claude/skills/project-scaffold/scripts/scaffold.py) that performs broad system-level operations.
  • [Privilege Escalation] (MEDIUM): Troubleshooting steps explicitly direct the user to use chmod +x to manually elevate script execution permissions.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 15, 2026, 09:30 PM