The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a webhook route handler in references/webhook-route.ts that ingests message content from external WhatsApp users. This constitutes an untrusted data ingestion point that could be used for indirect prompt injection attacks if an AI agent is configured to automatically process or respond to these messages. Ingestion points: The POST handler in references/webhook-route.ts. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are included in the reference implementation. Capability inventory: The skill possesses network communication capabilities via the wapi client in references/wapi-client.ts. Sanitization: The reference handler logs inbound text without specific sanitization, representing a standard risk factor for data-processing skills.
[DATA_EXFILTRATION]: The wapi client in references/wapi-client.ts performs network operations targeting api.wapi.crafter.run. These operations are required for the skill's primary messaging functionality and target the official infrastructure of the service provider, crafter-station. The implementation correctly scopes these operations to server-side environments to protect session credentials.

wapi-nextjs