telnyx-import-retell

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md shows the import endpoint (POST https://api.telnyx.com/v2/ai/assistants/import) pulls agent configurations directly from Retell and “Instructions” are imported as-is from Retell (a third‑party, user-generated source), so untrusted external content can influence assistant behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The curl/SDK call to https://api.telnyx.com/v2/ai/assistants/import is used at runtime to pull Retell agent configurations (imported "as‑is"), so external prompt/instruction content is fetched during execution and directly controls the assistant.