brainstorming

Warn

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: MEDIUMOBFUSCATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [OBFUSCATION]: The skill documentation contains a hidden control character (U+0007, BEL) embedded immediately before the name of the orchestrated skill autonomous-red-teamer in the "Database, ORM & Data" section of the integration matrix.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates an interactive ideation process where the agent solicitates and processes user-provided project requirements to generate architectural documents.
  • Ingestion points: User input is solicited during the "Product Ideation" phase (Step 2) and used to populate the "Understanding Lock" summary (Step 4).
  • Boundary markers: The skill lacks explicit delimiters or instructions to treat user input as untrusted data, potentially allowing malicious project descriptions to influence the agent's behavior during the design phase.
  • Capability inventory: The skill orchestrates multiple specialized tools with the ability to scrape web content (web-scraper), perform browser automation (browser-automation-expert), and write files to the local file system (PRD.md, BLUEPRINT.md).
  • Sanitization: No input validation or sanitization is performed on the user-provided ideation data before it is incorporated into the system design and delegated to other specialized skills.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 20, 2026, 01:29 PM
Security Audit — agent-trust-hub — brainstorming