Skills
skills/membranedev/application-skills/pymetrics/Gen Agent Trust Hub

pymetrics

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the membrane CLI to perform operations, requiring the agent to execute shell commands for login, connection management, and action execution.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @membranehq/cli package from NPM. This is a vendor-owned tool provided by the author (membranedev).
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it ingests and processes data from external Pymetrics API actions.
  • Ingestion points: Data returned from membrane action run commands.
  • Boundary markers: None specified in the instructions to distinguish external data from agent instructions.
  • Capability inventory: Full terminal command execution via the CLI.
  • Sanitization: No sanitization or validation of the retrieved Pymetrics data is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 11:35 PM