idea-trial

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (the 'case') and passes it through a series of automated subagent prompts including Prosecution, Defense, Jurors, and Judge. This structure creates a risk where malicious instructions embedded in the user's idea could attempt to influence or hijack the behavior of the subagents.
  • Ingestion points: User input for the business idea is captured in Section 1 and stored in the file 'idea-trial/00-case.md'.
  • Boundary markers: The skill does not utilize explicit delimiters or system-level instructions to separate the user-provided data from the instructions provided to the subagents.
  • Capability inventory: The skill uses the 'Task' tool to spawn parallel subagents and performs file read and write operations within the 'idea-trial/' directory to aggregate results.
  • Sanitization: No validation, escaping, or filtering is performed on the untrusted user input before it is interpolated into the prompts for the various subagents.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 10:03 AM