The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the latest GitHub CLI release and binary assets from official GitHub domains (api.github.com and github.com). This is documented in scripts/setup_gh.py which implements SHA256 verification and uses a secure extraction process (implementing filters to prevent path traversal). These sources are recognized as trusted, well-known services.
[COMMAND_EXECUTION]: Several scripts use the subprocess module to interact with the environment. For example, scripts/setup_gh.py executes git remote get-url origin to detect the repository slug, and scripts/github_project_setup.py executes the gh CLI to perform GraphQL queries. These commands are targeted and do not use shell execution for arbitrary input.
[DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file uses the exclamation-backtick syntax (!) to execute scripts/setup_gh.py --detect-only at load time. This command is used to auto-detect repository information and render usage examples into the agent's context. This operation is limited to local metadata detection and does not perform network calls in this mode.
[INDIRECT_PROMPT_INJECTION]: The skill processes backlog item data to create and manage GitHub issues in scripts/github_project_setup.py. While this presents an injection surface, the script uses the PyGithub library for structured API calls, providing inherent sanitization and boundary markers for external data.
[DYNAMIC_EXECUTION]: Test files such as tests/test_github_project_setup.py and tests/test_setup_gh.py use importlib to dynamically load and test the skill's scripts as modules. This is a standard pattern for automated testing in this context.

gh