piano

Warn

Audited by Gen Agent Trust Hub on Nov 15, 2026

Risk Level: MEDIUMMETADATA_POISONINGCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [METADATA_POISONING]: The skill's documentation contains conflicting information. The markdown body describes the skill as a "virtual piano" for musicians and links to Apple's MusicKit documentation, whereas the frontmatter and the actual CLI commands target the piano.io enterprise platform. This misleading metadata can cause confusion regarding the skill's true purpose and the destination of data.
  • [COMMAND_EXECUTION]: The skill relies on the execution of multiple membrane CLI commands to manage authentication, connection state, and API requests.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from the Piano.io API which may contain untrusted content. This content is subsequently interpreted by the agent, creating an injection risk.
  • Ingestion points: Data retrieved via membrane action run and membrane request (located in SKILL.md).
  • Boundary markers: No specific delimiters or instructions are provided to the agent to distinguish between platform data and system instructions.
  • Capability inventory: The agent can execute further shell commands and network requests based on data retrieved from these ingestion points.
  • Sanitization: There are no documented procedures for sanitizing or validating the response data before processing.
  • [EXTERNAL_DOWNLOADS]: The skill directs the user to install @membranehq/cli from the public npm registry. This is a vendor-provided tool required for the skill's operations.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 15, 2026, 05:40 AM