bbc-skill-tool

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via the runtime WhatsApp user messages that the bot routes to add_chatpdf (AI assistant instructions), i.e., the agent ingests arbitrary user-authored text not chosen by the operating user.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly calls update_answer with assistant.scrapeUrl (example: "https://www.business-website.com"), which will fetch website content at runtime and inject it into the add_chatpdf assistant’s knowledge/instructions, thereby directly controlling the agent's prompts.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)