The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions require the execution of shell commands to launch the application components, specifically npx electron . for the desktop UI and python3 .skills/skippy-voice/scripts/skippy-talk.py for the CLI.
[EXTERNAL_DOWNLOADS]: The skill depends on standard packages from official registries, including electron via npm and the elevenlabs SDK via PyPI.
[DATA_EXFILTRATION]: Microphone audio is captured and transmitted to ElevenLabs, a well-known technology service, to generate voice responses. This is the primary function of the skill and does not involve accessing sensitive local files.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes untrusted live audio data from the microphone. Malicious spoken instructions could potentially influence the agent's behavior through the returned transcriptions.
Ingestion points: app/index.html captures microphone data and sends it to the ElevenLabs WebSocket API.
Boundary markers: No specific delimiters or instructions are present in SKILL.md to help the agent distinguish between transcribed user speech and system commands.
Capability inventory: The skill is capable of executing shell commands for the Electron app and Python scripts as described in SKILL.md and README.md.
Sanitization: There is no evidence of sanitization or filtering of the text received from the voice-to-text service.

skippy-voice