skills/jeremylongshore/tons-of-skills-marketplace/firecrawl-architecture-variants/Gen Agent Trust Hub
firecrawl-architecture-variants
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes architecture patterns (on-demand scraping, scheduled crawls, and real-time pipelines) that ingest data from external URLs, creating a surface where untrusted content could influence downstream AI components.
- Ingestion points: External data is ingested through
firecrawl.scrapeUrl,firecrawl.asyncCrawlUrl, andfirecrawl.batchScrapeUrlsfunctions inSKILL.md. - Boundary markers: The provided code examples do not implement explicit boundary markers or delimiters to isolate untrusted web content from agent instructions.
- Capability inventory: The ingested content is subsequently processed for search indexing or stored in vector databases (
vectorStore.upsert) for use in AI/RAG pipelines. - Sanitization: The skill includes minimal content validation (checking string length and basic regex for CAPTCHAs) but lacks comprehensive sanitization or adversarial instruction filtering.
Audit Metadata