RedTeam

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands via curl to send status notifications to a local endpoint (http://localhost:8888/notify) in SKILL.md, Workflows/AdversarialValidation.md, and Workflows/ParallelAnalysis.md.- [PROMPT_INJECTION]: The workflow design presents an attack surface for indirect prompt injection by processing untrusted user content through a multi-agent pipeline.
  • Ingestion points: User-provided arguments and problem descriptions in Workflows/ParallelAnalysis.md and Workflows/AdversarialValidation.md are decomposed and passed to 32 parallel sub-agents.
  • Boundary markers: The agent prompt templates in Workflows/ParallelAnalysis.md use Markdown headers (e.g., ## THE ARGUMENT TO ANALYZE) and uppercase instruction blocks to delimit instructions from data.
  • Capability inventory: The skill possesses command execution capabilities (curl) for notification purposes.
  • Sanitization: No explicit sanitization or filtering of user-provided content is performed before it is interpolated into sub-agent prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 07:38 AM