The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/nim_call.py contains code that explicitly disables SSL certificate verification by setting ctx.check_hostname = False and ctx.verify_mode = ssl.CERT_NONE. This insecure configuration allows the NVIDIA_API_KEY and all processed data to be intercepted by malicious actors during transmission to NVIDIA's servers.
[PROMPT_INJECTION]: The skill facilitates indirect prompt injection by proxying untrusted user data to external language models.
Ingestion points: User-provided prompts are ingested via command-line arguments in scripts/nim_call.py and passed to the API.
Boundary markers: No delimiters or safety instructions are used to isolate user input within the API request payload.
Capability inventory: The script performs network operations to external APIs and returns the generated content to the user's terminal.
Sanitization: No input validation or output filtering is implemented to prevent malicious instructions from being executed by the external model or returned to the host agent.
[EXTERNAL_DOWNLOADS]: The skill communicates with the official NVIDIA NIM API endpoint at integrate.api.nvidia.com to perform inference tasks.

nim