The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on local shell commands including git, gh (GitHub CLI), and a custom gwt utility to perform its primary functions. These operations are restricted to the local development environment and are consistent with the skill's stated purpose of git worktree management.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where branch names are dynamically retrieved from GitHub PR metadata using the gh command.
Ingestion points: Branch names are ingested from remote GitHub PR data via gh pr view in SKILL.md.
Boundary markers: No explicit boundary markers or safety instructions are used to sanitize the ingested branch names.
Capability inventory: The skill executes shell commands (git fetch, gwt new, cd) using the retrieved branch names as arguments.
Sanitization: Examples in the documentation include shell quoting (e.g., "$BRANCH"), which follows standard practices for handling variables in shell scripts and reduces the risk of word-splitting attacks.
[DATA_EXPOSURE]: The skill's instructions involve copying .env and .dev.vars files to new worktrees. This is standard developer behavior for maintaining project configuration across different branch contexts and does not involve external exfiltration.

gwt