The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the creation and management of Lefthook configuration files (e.g., lefthook.yml), which define shell commands and scripts to be executed during Git lifecycle events such as pre-commit or pre-push.
[EXTERNAL_DOWNLOADS]: The skill documents the remotes and extends features, which allow Lefthook to download and merge configuration settings from external Git repositories or local files. It also provides instructions for installing the Lefthook binary using standard package managers including npm, pipx, RubyGems, and Go.
[INDIRECT_PROMPT_INJECTION]: The skill describes how to process external data sources such as staged file lists ({staged_files}) and Git commit messages ({1}), which represent an attack surface for indirect prompt injection if the tools invoked by the hooks (e.g., linters or commit validators) are vulnerable to malicious content within those inputs.
Ingestion points: File paths provided via {staged_files}, {push_files}, and {all_files} templates, and commit message content provided via positional placeholders like {1} in commit-msg hooks.
Boundary markers: The skill does not explicitly define boundary markers for these inputs, relying on the user's chosen hook commands (e.g., eslint, prettier) to handle the data safely.
Capability inventory: The skill is designed to manage subprocess execution via the run and scripts keys in Lefthook configurations.
Sanitization: No specific sanitization or escaping mechanisms are provided within the skill instructions for the interpolated Git metadata.

lefthook