scrapinghub

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Installs the @membranehq/cli package from the official NPM registry to provide the necessary tooling for platform interaction.
  • [COMMAND_EXECUTION]: Utilizes the membrane CLI to perform authentication, manage connections, and execute Scrapinghub actions via shell commands.
  • [DATA_EXFILTRATION]: Facilitates network communication with the Membrane platform's official infrastructure (getmembrane.com) to execute actions and manage credentials. No unauthorized data exfiltration patterns were identified.
  • [PROMPT_INJECTION]: As the skill processes data retrieved from web scraping (Scrapinghub), it inherently possesses an indirect prompt injection surface where scraped content could potentially contain instructions intended for the agent. (Ingestion points: membrane action run output; Boundary markers: none; Capability inventory: shell command execution; Sanitization: none).
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 02:47 AM
Security Audit — agent-trust-hub — scrapinghub