parseur

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Installs the @membranehq/cli package from the official npm registry. This is a vendor-provided tool used to facilitate integration with the Membrane platform.\n- [COMMAND_EXECUTION]: Executes several membrane shell commands to manage authentication, list connections, and run document parsing actions. These operations are within the scope of the skill's stated purpose.\n- [COMMAND_EXECUTION]: Uses npx to dynamically execute the Membrane CLI for action discovery, which is a standard method for running Node-based tools without permanent installation.\n- [PROMPT_INJECTION]: The skill processes untrusted natural language data via the intent parameter in membrane action list within SKILL.md. This data is used to discover and subsequently execute actions (membrane action run), creating a capability chain where external instructions could potentially influence tool selection. No explicit boundary markers or sanitization logic are defined in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 04:04 PM
Security Audit — agent-trust-hub — parseur