sales-page

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: All core functionalities and reference materials are benign and focused on marketing tasks.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection when using WebFetch to analyze external competitor sales pages. Ingestion points: Competitor data retrieved via WebFetch in SKILL.md. Boundary markers: None identified between external data and agent instructions. Capability inventory: Read, Write, Edit, WebSearch, and WebFetch as listed in SKILL.md. Sanitization: None performed on retrieved web content. This surface is considered safe/low risk given the specific marketing context.
  • [EXTERNAL_DOWNLOADS]: Research-based web fetching is performed through WebSearch and WebFetch. No execution of remote scripts or binaries occurs.
  • [COMMAND_EXECUTION]: No unauthorized commands or subprocesses are executed. File operations are limited to the provided toolset for blueprint creation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 10:41 PM