The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command syntax to automatically execute lazyworktree list --json, pwd, and git branch --show-current when the skill is loaded. These are benign commands used to provide the agent with immediate context about the current working environment and repository state.\n- [COMMAND_EXECUTION]: The skill is designed to execute shell commands within specific worktrees using the lazyworktree exec command and the --exec post-creation hook. This functionality is intended for its primary purpose of managing development tasks across multiple worktrees.\n- [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting untrusted data from external sources.\n
Ingestion points: The skill fetches information from GitHub/GitLab PRs and issues via the --from-pr and --from-issue flags.\n
Boundary markers: There are no explicit delimiters or instructions to the agent to ignore potentially malicious instructions embedded within the PR or issue descriptions.\n
Capability inventory: The skill possesses extensive command execution capabilities via lazyworktree exec.\n
Sanitization: The documentation states that worktree names are automatically sanitized (replacing special characters with hyphens), which reduces the risk of path-based command injection during name generation.

worktree-manager