competitor-profiling
Pass
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill performs network operations to external, non-whitelisted domains as part of its research process.
- The skill uses
firecrawlanddataforseotools to access arbitrary competitor URLs provided by the user to extract positioning, features, and pricing information. - It also accesses well-known review platforms such as G2 and Capterra to gather customer sentiment data.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from untrusted third-party websites.
- Ingestion points: Untrusted data enters the agent context through
firecrawl_scrapecalls to arbitrary competitor URLs and review pages (SKILL.md). - Boundary markers: The skill implements a process where raw scraped data is saved to a local
competitor-profiles/raw/directory before being read for synthesis, providing a point for potential audit, though it does not automatically prevent injection. - Capability inventory: The skill has the capability to write files to the local filesystem (
competitor-profiles/) and use network tools to reach external sites. - Sanitization: There are no explicit instructions within the skill to sanitize, filter, or ignore potential instructions embedded in the scraped markdown content before the agent synthesizes the final profile.
Audit Metadata