ableton-live
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references the AbletonOSC repository on GitHub and suggests installing common libraries from PyPI (python-osc) and NPM (osc-js).
- [COMMAND_EXECUTION]: Provides instructions for running the AbletonOSC server via the command line and installing required packages using standard package managers.
- [PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection via the processing of untrusted external data.
- Ingestion points: The documentation includes code for listening to network-based OSC messages and reading state data from local JSON files.
- Boundary markers: No delimiters or isolation instructions are present to prevent the agent from interpreting embedded commands in ingested data.
- Capability inventory: The skill utilizes network communication (UDP) and local file system access for status persistence.
- Sanitization: There is no evidence of validation or sanitization for incoming OSC message arguments or data loaded from temporary files.
Audit Metadata