The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: Executes an installation script from the vendor's official domain (vapi.ai) via a piped shell command. It also includes instructions for running a remote MCP server via npx using the vendor's official documentation endpoint.
[EXTERNAL_DOWNLOADS]: Fetches configuration files and setup scripts from official vendor domains including vapi.ai and docs.vapi.ai.
[COMMAND_EXECUTION]: Utilizes curl to interact with the vendor's API endpoints (api.vapi.ai) for patching assistant settings and phone number configurations.
[PROMPT_INJECTION]: The skill sets up infrastructure to ingest and process real-time transcripts and tool calls from external phone interactions, which introduces a surface for indirect prompt injection.
Ingestion points: The webhook server receives transcript, tool-calls, and assistant-request messages containing external user data.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the provided implementation examples.
Capability inventory: The skill allows the agent to execute shell commands (curl) and modify remote service configurations.
Sanitization: The example code demonstrates processing incoming JSON payloads without explicit sanitization or validation of the text content within transcripts.

setup-webhook