command-development

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill demonstrates how to create commands that ingest user-provided arguments ($1, $ARGUMENTS) and file contents (@file) directly into prompts without boundary markers or sanitization. This creates a surface for indirect prompt injection where malicious content in a file or argument could influence the agent's behavior.
  • Ingestion points: $1, $2, $ARGUMENTS, @$1 in examples/simple-commands.md.
  • Boundary markers: Absent in examples.
  • Capability inventory: Bash (git, npm, kubectl, helm), Read, Write.
  • Sanitization: Absent.
  • Command Execution (SAFE): The skill documents the use of bash execution via backticks (!...). While powerful, the documentation explicitly recommends using restrictive 'allowed-tools' configurations (e.g., Bash(git:*)) to mitigate risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:21 PM