chatbot-whatsapp
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow that ingests untrusted text from WhatsApp messages (e.g.,
{{1.value.messages[].text.body}}) and interpolates it directly into LLM prompts within automation platforms. - Ingestion points: WhatsApp Business API messages received via Meta webhooks.
- Boundary markers: The skill provides basic system prompt instructions but does not specify robust delimiters or defensive instructions to prevent the model from following adversarial user input.
- Capability inventory: The automation scenarios include HTTP request capabilities (interacting with Meta's messaging API) and integrations with Google Sheets.
- Sanitization: No input validation or sanitization steps are documented for the incoming user message before it is processed by the AI.
Audit Metadata