The Agent Skills Directory

NO_CODE (SAFE): The skill is composed entirely of Markdown documentation files. It contains no executable scripts, binary files, or configuration files that would be automatically executed by an environment.
COMMAND_EXECUTION (SAFE): All command-line examples provided (e.g., gh api, curl, jq) are standard tools for interacting with the GitHub API. The shell snippets for automation, such as cleaning up artifacts or merging PRs, are legitimate administrative scripts intended for use by an authorized agent.
CREDENTIALS_UNSAFE (SAFE): The skill does not contain any hardcoded secrets. It provides clear instructions on authentication and uses placeholders like YOUR_TOKEN, INSTALLATION_TOKEN, and SECRET_NAME for user-provided credentials.
Indirect Prompt Injection (SAFE): Although the skill involves reading data from external sources such as GitHub issues and pull requests, it does not include logic for the automated execution of untrusted content. It serves as a manual or agent reference for performing authorized API operations.

github-api