harvest-deep-crawl

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads content from arbitrary external URLs provided by the user during the crawl process.
  • [COMMAND_EXECUTION]: Utilizes the Bash tool to execute curl commands against a local service (localhost:11235) to manage the crawling engine.
  • [PROMPT_INJECTION]: The skill processes untrusted external content, which presents a surface for indirect prompt injection.
  • Ingestion points: Content retrieved from external websites via WebFetch or the local crawler.
  • Boundary markers: The skill does not explicitly define delimiters to separate crawled content from instruction context.
  • Capability inventory: The skill utilizes Write permissions to cache content and Bash for tool interaction.
  • Sanitization: There is no evidence of sanitization or filtering of the crawled web content before it is stored and provided to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 07:46 AM
Security Audit — agent-trust-hub — harvest-deep-crawl