neurokit2
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install its development version from a remote GitHub repository at https://github.com/neuropsychology/NeuroKit/zipball/dev.
- [INDIRECT_PROMPT_INJECTION]: The skill processes physiological signals (ECG, EEG, EDA, RSP, EMG, EOG) which could serve as a vector for indirect prompt injection if the data or its associated metadata contains malicious instructions.
- Ingestion points: Biosignal data is ingested through signals provided as arrays or dataframes to functions such as
nk.ecg_process()ornk.eeg_power(). - Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore potential commands embedded in the processed data.
- Capability inventory: The skill allows the agent to process signals, compute complex metrics, and generate visualizations using matplotlib.
- Sanitization: While the skill applies mathematical filters for signal cleaning (noise reduction), it does not sanitize the input data for natural language prompt injection patterns.
Audit Metadata