atelier-oracle-challenge

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes user-provided data through the $ARGUMENTS variable without using boundary markers or protective instructions.
  • Ingestion points: The $ARGUMENTS variable in SKILL.md acts as an entry point for untrusted data provided during the /atelier-challenge command invocation.
  • Boundary markers: The skill fails to use delimiters (such as XML tags or triple quotes) to separate the untrusted input from the system instructions.
  • Capability inventory: The skill leverages the mcp__sequential-thinking__sequentialthinking tool to perform multi-step analysis based on the input.
  • Sanitization: There is no evidence of input validation or instructions telling the agent to ignore any commands that might be embedded within the $ARGUMENTS string.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 10:18 AM