The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The skill requires the installation of the @kapso/whatsapp-cloud-api package via npm. This package is not provided by a trusted organization or repository listed in the security guidelines.
COMMAND_EXECUTION (MEDIUM): The documentation instructs the user/agent to execute multiple local JavaScript scripts (e.g., list-platform-phone-numbers.mjs, create-template.mjs, send-template.mjs) using Node.js. These scripts are intended to interact with the Kapso API and manage local assets.
DATA_EXFILTRATION (LOW): The skill transmits data and credentials (KAPSO_API_KEY) to api.kapso.ai. This is the documented purpose of the skill (Meta proxy), but it involves sending potentially sensitive messaging data to a non-whitelisted third-party domain.
PROMPT_INJECTION (LOW): The skill is vulnerable to Indirect Prompt Injection (Category 8) because it reads external data from WhatsApp inboxes.
Ingestion points: Incoming messages are retrieved via client.messages.query(), client.conversations.list(), or direct proxy GET requests to the /messages endpoint.
Boundary markers: None identified in the provided instructions; untrusted message content is processed directly into the agent's context.
Capability inventory: The skill can send outbound messages, create/modify templates, and upload media, which could be leveraged if an attacker successfully injects instructions into an incoming message.
Sanitization: No sanitization or validation of the message content is mentioned before it is processed by the agent.

whatsapp-messaging