tavily-crawl

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions for executing shell commands using the tvly (Tavily CLI) tool, including tvly crawl, tvly login, and tvly --status. It also suggests installing the tool via uv tool install or pip install.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates downloading web content from external domains through the Tavily service. It includes specific warnings and filters (e.g., --select-paths, --exclude-domains) to limit the scope of these downloads, but the primary function is network-based data collection.
  • [PROMPT_INJECTION]: The skill identifies the risk of indirect prompt injection in the 'Anti-Patterns' section, warning against treating crawled pages as trusted instructions. It includes a mandatory evidence chain for verification (Ingestion points: tvly crawl output; Boundary markers: --instructions and --chunks-per-source; Capability inventory: File writing via --output-dir and network access; Sanitization: Explicitly warns to treat results as untrusted data).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 01:38 AM
Security Audit — agent-trust-hub — tavily-crawl