renoise-cli

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEMETADATA_POISONINGEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [METADATA_POISONING]: The author field in SKILL.md is set to 'renoise', which differs from the provided author context of 'arcocodes'.
  • [EXTERNAL_DOWNLOADS]: The skill relies on external scripts located in a relative sibling directory ('../renoise-setup/') to install and verify the 'renoise' binary, which is a core dependency not contained within the skill itself.
  • [COMMAND_EXECUTION]: Automated preflight checks execute shell commands to modify the system environment ('PATH') and run local utility scripts using node, bash, and python.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through user-provided prompts and media files used for video generation. While the skill relies on server-side validation, the ingestion of untrusted data constitutes an attack surface.
  • Ingestion points: Prompts are read from project JSON files and media analysis is performed on local files.
  • Boundary markers: The skill uses '--prompt-file' to handle input via files or stdin, which reduces shell injection risks compared to direct arguments.
  • Capability inventory: The skill can read/write local files and execute the renoise CLI for network uploads.
  • Sanitization: Relies on the external CLI and server-side API for content moderation and safety filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 04:38 PM
Security Audit — agent-trust-hub — renoise-cli