meta-prompting

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill is composed of purely instructional text for the AI agent. No code, scripts, or system-level operations are included.
  • [NO_CODE]: This skill contains no scripts, binaries, or configuration files that execute logic; it is entirely composed of natural language instructions.
  • [DATA_EXFILTRATION]: No network access or file reading capabilities are defined in the skill.
  • [PROMPT_INJECTION]: While the skill uses terms like '/constrain' or 'Override', these are part of the intended functionality to manage the agent's output format and length, not attempts to bypass security filters.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 02:12 PM
Security Audit — agent-trust-hub — meta-prompting