telnyx-voice-go

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

• Third-party content exposure detected (high risk: 0.75). The required runtime path includes Telnyx webhook handling where the agent reads body from an inbound HTTP request (io.ReadAll(r.Body)) and parses it into an event object (client.Webhooks.Unwrap(body, r.Header); webhook payloads are outsider-authored free text (originating from Telnyx/caller/SIP) that can contain attacker-controlled content.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I reviewed the skill docs for literal, high-entropy credentials. Most values are placeholders or simple examples (environment variable usage TELNYX_API_KEY, masked IDs like "7267xxxxxxxxxxxxxx", example E.164 numbers, numeric IDs) which are not secrets per your rules.

However, the string "v3:MdI91X4lWFEs7IgbBEOT9M4AigoY08M0WWZFISt1Yw2axZ_IiE4pqg" is a long, random-looking token (base64url-like) and appears as a CallControlID value. The documentation describes CallControlId as a "unique identifier and token for controlling the call", so this literal likely represents a real, usable credential/token. It meets the high-entropy definition and is not a placeholder, so I flag it.