The Agent Skills Directory

[SAFE]: The skill's primary function is audio synthesis, and its operations are consistent with this purpose. It does not exhibit signs of credential theft, unauthorized data access, or malicious persistence.
[COMMAND_EXECUTION]: The skill generates and runs localized Python scripts to perform DSP tasks. These scripts rely on well-known libraries like NumPy and SciPy. The execution is handled via 'uv', which manages dependencies from official registries.
[EXTERNAL_DOWNLOADS]: The skill uses research tools to identify environmental characteristics. This data is used as parameters for the synthesis engine. No untrusted scripts or binaries are downloaded.
[PROMPT_INJECTION]: The skill's ingestion of web data is limited to technical research. There is no evidence of vulnerable prompt interpolation that could lead to indirect injection. Mandatory Evidence Chain: (1) Ingestion points: WebFetch in Step 2. (2) Boundary markers: Research results are distilled into specific technical attributes. (3) Capability inventory: Subprocess calls via uv run and file writes in Step 4/6. (4) Sanitization: Synthesis logic uses parsed parameters rather than direct output interpolation.

create-soundscape