The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the git and gh (GitHub) CLI tools to perform its core operations. Command construction is carefully controlled, using validated and normalized inputs (such as repository targets and work item slugs) to ensure that sub-processes are restricted to authorized actions within the target repository.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external content from GitHub issue bodies and comments. It treats this information as untrusted by default, implementing validation logic and using specific fences (e.g., ) to isolate and verify structured data before it is used for decision-making.
[SAFE]: The skill demonstrates a strong security posture. It protects sensitive environment tokens by filtering the environment of sub-processes, employs atomic write patterns to ensure data integrity, and strictly validates all paths and identifiers to prevent directory traversal or unauthorized file access.

ticket