atelier-orchestrator

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill employs aggressive, mandatory language designed to override agent decision-making and autonomy regarding the use of other skills.
  • Evidence: Use of phrases like "YOU ABSOLUTELY MUST invoke the skill", "YOU DO NOT HAVE A CHOICE", "This is not negotiable", and "You cannot rationalize your way out of this" in the <EXTREMELY-IMPORTANT> block.
  • [PROMPT_INJECTION]: The skill includes a "Red Flags" section that instructs the agent to disregard internal reasoning or perceived user urgency in favor of the skill's rigid workflow.
  • Evidence: The "Red Flags — You Are Rationalizing" table explicitly counters natural agent heuristics (e.g., "The human seems impatient", "I'll just do a quick prototype", "I already know how to do this") with absolute directives to follow the workflow regardless of context.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 04:29 PM
Security Audit — agent-trust-hub — atelier-orchestrator