The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands using git and gh (GitHub CLI) to manage branches, worktrees, issues, and pull requests. These operations are essential for the skill's primary function and are driven by the user's workflow commands.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection by reading and processing content from external or local sources without adequate defenses.
Ingestion points: In SKILL.md, the update and review code commands ingest data from GitHub issue bodies (retrieved via gh issue view) and from local plan files located in .claude/plans/ or docs/plans/.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to the agent to disregard potential commands or role-play attempts within the ingested text.
Capability inventory: The skill can modify the local file system through git worktree, change the state of remote GitHub repositories (editing issues, PRs, and labels), and interact with other AI agents via /ask codex.
Sanitization: Absent. Content from issues and plans is interpolated directly into outgoing prompts or command arguments.
[DATA_EXFILTRATION]: The skill transmits local project information, including code diffs and development plans, to GitHub repositories to create or update issues and pull requests. This data transfer is a core component of the intended workflow but represents an outbound movement of local project data.

issue-driven