firecrawl-cost-tuning
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected in the skill content.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data from web sources via Firecrawl. 1. Ingestion points: Data is received from the Firecrawl API (SKILL.md). 2. Boundary markers: The 'Output' section explicitly instructs: 'Do not include credentials, page contents, or customer data.' 3. Capability inventory: The skill uses Read and Grep tools and demonstrates network operations via curl and the firecrawl-js library. 4. Sanitization: The skill focuses on metadata and cost metrics rather than parsing page content, reducing the need for content-specific sanitization in these examples.
Audit Metadata