The Agent Skills Directory

[SAFE]: The skill is purely informational, providing guidance on managing the context window and token limits for better performance. No executable code or suspicious patterns were found.
[SAFE]: No external network requests, data exfiltration, or hardcoded credential exposure were detected.
[SAFE]: No prompt injection or behavior override attempts are present in the documentation content.
[SAFE]: The skill references standard built-in commands like /compact and /cost for educational purposes without attempting to manipulate the agent's core safety logic.

context-budgeting