tgravity-work-prompt-architect

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security risks such as credential exfiltration, remote code execution, or privilege escalation were detected.
  • [DATA_EXPOSURE]: The skill implements safety constraints in its instructions, explicitly forbidding the use of sensitive data like employee privacy, official seals, or contract scans in generated examples.
  • [INDIRECT_PROMPT_INJECTION]: Although the skill is designed to process untrusted prompts for auditing, it lacks any dangerous capabilities (e.g., command execution, file writing, network calls) that would enable an injection to impact the underlying system.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 02:44 AM
Security Audit — agent-trust-hub — tgravity-work-prompt-architect