S2-SP-OS Vision Cast

S2-SP-OS Vision Cast. Features a universal Protocol Sniffer (AirPlay/Chromecast/DLNA) for native casting, backed by our secure S2 ephemeral push fallback. /...

This skill appears coherent for LAN casting: it probes local casting ports and can fall back to an S2 push mechanism. Before installing, verify the following: (1) Source/trust — the package homepage and author are unknown; confirm the vendor and review any externally hosted receiver/service docs. (2) Token scope — find out what S2_VISION_TOKEN can do (which endpoints it authorizes and whether it allows arbitrary pushes to user devices). Limit the token's privileges if possible. (3) Network behavior — the visible code shows only local port probes and a stubbed push; ask the author for the real push implementation and endpoints so you can review where payloads are sent. (4) Consent is only an env var — S2_PRIVACY_CONSENT=1 is a simple guard; ensure agents set it intentionally and verify local policy before allowing scans on shared networks. If you need higher assurance, run the script in a controlled test environment and inspect traffic (e.g., with a packet capture) before deploying on production networks.

Type: OpenClaw Skill Name: s2-vision-projection Version: 1.1.0 The skill bundle provides a framework for discovering and interacting with media casting devices (AirPlay, Chromecast, DLNA) on a local network. The core logic in `vision_cast.py` implements a port-knocking 'sniffer' that is explicitly restricted to private/LAN IP addresses to prevent SSRF or external scanning. The tool requires explicit user consent via environment variables (S2_PRIVACY_CONSENT) and follows a well-documented workflow in SKILL.md and AGENT-EXAMPLES.md that prioritizes native protocols and privacy. No indicators of data exfiltration, unauthorized execution, or malicious prompt injection were identified.