firecrawl-lead-gen
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill extracts data from external websites and prospect databases, which are untrusted sources. This data is then processed to create lead lists. If these external sources contain malicious instructions designed to influence the agent's behavior, it could lead to indirect prompt injection vulnerabilities during the extraction and summarization phases.
- Ingestion points: Data extracted from web directories and prospect databases via the Firecrawl browser as described in
SKILL.md. - Boundary markers: The instructions lack explicit delimitation or 'ignore embedded instructions' directives for the extracted web content, although they do specify a
Quality Barfor legitimate extraction. - Capability inventory: The agent uses Firecrawl for web navigation and data extraction, and generates structured output (JSON/CSV/Markdown) which involves processing potentially adversarial text.
- Sanitization: No explicit sanitization or validation of the extracted content is mentioned before it is interpolated into the final deliverable markdown or data files.
Audit Metadata