telegram-bot-builder-hardened

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly ingests untrusted, user-generated Telegram updates (messages and media) via "Webhook Integration" and "Auto-reply / Filters" described in SKILL.md and demonstrated in SAFETY.md (e.g., examples that download and process photos/videos and handle incoming updates), so third‑party content sent by arbitrary users can be read and materially influence bot behavior.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly includes payment functionality and payment-specific operations. It lists "Payment (Stars)" in features and the guardrails reference generating invoice code and handling provider tokens/payment credentials. Those are specific payment-related capabilities (invoice generation and provider tokens) rather than generic tooling, so this skill provides direct financial execution functionality.