The Agent Skills Directory

Documentation Only: The skill consists purely of markdown documentation describing system architecture and does not include any scripts, commands, or executable logic.
Security Architecture Guidance: The content proactively addresses security concerns, such as process isolation (Electron Main vs. Renderer), workspace trust requirements for local file access, and the use of isolated-world preload scripts to mitigate risks from hostile pages.
Injection Risk Awareness: The documentation explicitly highlights that page content should be treated as untrusted model input, specifically mentioning potential prompt injection risks and the need for privacy gates and network filters.

integrated-browser