The Agent Skills Directory

[SAFE]: The skill gathers project context by reading local files or fetching content from user-provided URLs and issue trackers. This activity is restricted to the primary purpose of requirement analysis.
[SAFE]: File system operations are limited to reading the current codebase for domain knowledge and writing tickets to a local .scratch directory, which is standard behavior for development-oriented tools.
[SAFE]: The integration with external trackers (GitHub, Linear) occurs only after a structured review and approval step by the user, providing a significant safeguard against automated exploitation or indirect injection.
[SAFE]: Security hardening is present via frontmatter configurations (disable-model-invocation: true and allow_implicit_invocation: false), which limit the agent's autonomous behavior.

to-tickets