The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on system commands (git, gh) to perform its core functions, such as creating worktrees (git worktree add), managing branches, and modifying GitHub issue metadata (gh issue edit). While these are intended functionalities for a development tool, they involve direct system interaction based on programmatically generated strings derived from external data.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface. 1. Ingestion points: Untrusted data enters the agent's context through issue body retrieval via the gh issue view command in the /issue update and /issue worktree subcommands. 2. Boundary markers: No explicit delimiters or instructions are used to separate issue content from system instructions, meaning the agent might treat content within an issue body as a command. 3. Capability inventory: The skill can perform file system operations (worktree creation), modify remote repository state (PR and issue edits), and call internal agent tools (code review via Codex). 4. Sanitization: No sanitization or validation of the retrieved issue body is performed before processing, allowing potentially malicious instructions to influence the agent's behavior.

issue-driven