The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (HIGH): The skill documentation and the upload_audio function in scripts/voice_call.py explicitly use https://tmpfiles.org/api/v1/upload. This domain is flagged as a Botnet-related URL by security scanners.
[DATA_EXFILTRATION] (HIGH): User-generated audio content is uploaded to an unauthenticated, flagged third-party service (tmpfiles.org), risking exposure of sensitive communication.
[COMMAND_EXECUTION] (LOW): The Python script executes curl and twilio CLI commands via subprocess.run. While it uses list-based arguments to mitigate shell injection, it relies on the security and availability of external binaries.
[CREDENTIALS_UNSAFE] (LOW): The skill requires an ELEVENLABS_API_KEY and suggests storing it in a .env file, which requires careful local management to prevent accidental exposure.
[PROMPT_INJECTION] (LOW): (Category 8) This skill presents a surface for indirect prompt injection via the args.message input. 1. Ingestion point: args.message in scripts/voice_call.py. 2. Boundary markers: Absent. 3. Capability inventory: subprocess.run calls to curl and twilio. 4. Sanitization: Absent.

twilio-phone