The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (python3 scripts/call_user.py) to facilitate voice interactions and approval workflows.
[EXTERNAL_DOWNLOADS]: The documentation in references/setup.md suggests using npx @parcha/hands-free install for setup, which downloads and executes external code from the npm registry.
[DATA_EXFILTRATION]: The scripts/call_user.py script sends data (agent messages and Vapi credentials) to api.vapi.ai. While necessary for the skill's purpose, this involves sharing context with a third-party service provider.
[PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by converting external audio input into agent-contextual text.
Ingestion points: User speech is captured via the Vapi API and output by scripts/call_user.py in SKILL.md workflows.
Boundary markers: Absent. The agent is explicitly told to treat the transcript as direct user input.
Capability inventory: The agent possesses full local development capabilities, including command execution and file modification.
Sanitization: No input validation or sanitization is performed on the transcripts before they are returned to the agent.

hands-free