setup-inbound

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

• Third-party content exposure detected (high risk: 1.00). Yes — the skill's caller-identification flow (references/caller-identification.md and SKILL.md) shows Bolna fetching third-party sources (public Google Sheets and arbitrary HTTP GET endpoints / CSV uploads) and merging that untrusted, user-provided data into recipient_data which the agent must read/use in prompts and for call-routing/allow/deny decisions, enabling indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). Bolna performs runtime fetches to external caller-identification sources — e.g. your Internal API endpoint like https://api.your-domain.com/customers and publicly-shared Google Sheet URLs — and merges the returned JSON/CSV into recipient_data that directly controls agent prompt variables, so these external URLs are runtime dependencies that can influence agent instructions.