competitor-analysis
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates entirely within its documented scope of competitive research and analysis. It uses standard web search and scraping tools to gather information and returns structured JSON output to the user.
- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests untrusted data from the internet; however, its design for structured extraction mitigates common risks.
- Ingestion points: The skill retrieves content from competitor homepages, pricing tables, and product documentation (referenced in SKILL.md strategy section).
- Boundary markers: The instructions do not explicitly require the use of boundary markers or instructions to ignore embedded commands in the scraped content.
- Capability inventory: Capabilities are limited to web searching, scraping, and spawning worker agents for parallel analysis. No file-system modification or sensitive network operations are requested.
- Sanitization: There is no specific requirement for the agent to sanitize or filter the raw scraped text before processing it for summary and matrix generation.
Audit Metadata