amc-run-rtsp-calibration

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly requests RTSP URLs (which may embed username:password) and an optional VIOS bearer token, and shows examples/env-vars containing embedded credentials and JSON payloads that would be sent to the service—so an agent could be asked to collect and then verbatim-embed those secrets into requests or displayed commands if not careful.

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/run_rtsp_calibration.py, the workflow ingests free text supplied by the outsider at runtime via STREAMS_JSON/RTSP_URLS (parsed in _parse_streams) and then sends that data in capture_body to POST {BASE_URL}/rtsp/capture/{project_id} (Step 2), so untrusted text can be posted into the pipeline without selecting a specific item first.