guitar-dsp
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied WAV audio files and RTNeural JSON model files through its utility scripts (e.g., alias_probe_report.py, model_package_summary.py). This creates a surface for indirect instructions to be processed by the agent when analyzing user-provided data.
- Ingestion points: scripts/model_package_summary.py (JSON), scripts/receptive_field.py (JSON), scripts/alias_probe_report.py (WAV), scripts/compare_audio_metrics.py (WAV).
- Boundary markers: The skill emphasizes strict separation between data loading/parsing (non-realtime) and audio processing (realtime), advising that heavy operations stay off the audio thread.
- Capability inventory: The utility scripts calculate and print audio metrics (ESR, RMSE, aliasing energy) using standard math libraries. The skill itself is designed to guide the generation of C++/JUCE DSP code.
- Sanitization: The scripts utilize standard Python library parsers (json, wave) which provide basic structural validation of input files, but do not perform deep content sanitization against adversarial data.
Audit Metadata