The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes git and gh (GitHub CLI) to perform repository and issue management tasks. These commands are scoped to the skill's primary function (e.g., gh issue create, gh repo view, git rev-parse).
[INDIRECT_PROMPT_INJECTION]: The skill processes data from GitHub issues and comments which are potentially attacker-controlled. However, it mitigates this risk by employing strict Markdown templates for parsing and providing explicit instructions to the agent to treat issue content as data rather than instructions, preventing the accidental execution of embedded shell commands.
[EXTERNAL_DOWNLOADS]: References are made to an external script (scripts/plan-header.mjs) associated with a separate /execute skill for computing integrity hashes. This is a local cross-skill dependency and does not involve unverified remote downloads or remote code execution.

issue