The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a 'Lifecycle hooks' feature that executes arbitrary shell commands defined in configuration files.
Evidence found in scripts/lib/util.sh (function run_hook) and scripts/lib/workspace.sh (function run_ws_hook).
Commands are executed via bash -c "$hook_cmd".
Hook definitions (preCreate, postCreate, preDelete, postDelete) are read from repository-specific files such as .super-worktree.json or super-worktree.workspace.json.
[CREDENTIALS_UNSAFE]: The skill is explicitly designed to discover and propagate sensitive files across the file system.
The copy_sensitive_files function in scripts/lib/sync.sh targets patterns like .env, *.secret, *.key, credentials.json, and auth.json.
The configuration allows for environment variable interpolation in copyFiles paths, enabling the copying of global secrets such as ${HOME}/.aws/credentials into project worktrees.
[EXTERNAL_DOWNLOADS]: The script scripts/validate-skills.sh performs dynamic package installation.
It executes npm install --silent yaml@^2.8.0 to a temporary directory if the package is not already present.

super-worktree