The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) and its API interface to perform repository operations such as creating, viewing, and editing issues and pull requests.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from GitHub issue bodies and comments. It includes explicit defensive instructions to never derive authority or instructions from the text of issues, effectively mitigating risks associated with malicious instructions embedded in external content.
[DATA_EXPOSURE]: Before transmitting data to the target host, the skill performs a privacy scan of the title and body content. It is instructed to identify and redact sensitive information including local paths, usernames, hostnames, and credentials, replacing them with generic placeholders like <token> or <hostname>.
[PRIVILEGE_ESCALATION]: The skill implements a strict capability check. It verifies the authenticated user's viewerPermission (e.g., MAINTAIN or ADMIN) on the target host before attempting to apply protected labels or approve workflows, preventing unauthorized actions by users with insufficient privileges.
[DATA_HANDLING]: The skill uses temporary files with restricted permissions (0700 or 0600) to handle data and ensures these files are cleaned up immediately after execution, minimizing the risk of local data leakage.

issue-creation