The Agent Skills Directory

[SAFE]: The skill incorporates specific defensive instructions to prevent the agent from autonomously executing network-sensitive discovery or setup commands, even when prompted by simulated urgency or external authority.
[SAFE]: Includes explicit data protection rules that forbid the inclusion of Bridge IP addresses, API keys, or device serial numbers in agent responses, mitigating social engineering risks.
[SAFE]: Implements exfiltration prevention by explicitly prohibiting the transmission of local device information to network-facing endpoints or via commands like scp.
[SAFE]: Installation and dependencies are managed through established channels, specifically a Homebrew formula for the OpenHue CLI tool.
[SAFE]: The skill demonstrates a high level of security awareness by documenting safety evaluations and providing clear boundaries for agent behavior when interacting with local infrastructure data.

openhue-hardened