whatsapp-bot-builder
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents a surface for processing untrusted user input via webhooks, which could be used for prompt injection attacks.
- Ingestion points: Incoming messages are processed via the Express.js POST endpoint at
/webhookdefined in SKILL.md. - Boundary markers: Not present in the provided tutorial code snippets.
- Capability inventory: The skill demonstrates the ability to send messages, media, and interactive components via the Meta Graph API using fetch.
- Sanitization: No explicit validation or filtering of the message content is implemented in the provided examples.
- [SAFE]: All external URLs point to official Meta/Facebook developer and business platforms, which are recognized well-known services.
- [SAFE]: The skill correctly instructs users to store sensitive credentials like access tokens in environment variables and enforces the use of HTTPS for webhooks.
Audit Metadata