The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Automated scanners (URLite) flagged the documentation link https://jeffallan.github.io/claude-skills/skills/devops/cli-developer/. This URL belongs to the skill's author and is hosted on GitHub Pages, a well-known service. The alert may be a false positive resulting from automated categorization, but users should be aware of the external link.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the development of CLI tools based on user specifications, which creates a surface for indirect prompt injection if the user input contains malicious instructions meant to manipulate the generated code.
Ingestion points: CLI command names, arguments, and descriptions provided by the user during the implementation workflow (SKILL.md).
Boundary markers: No explicit delimiters or boundary markers are defined to isolate user-provided data from the agent's internal instructions.
Capability inventory: The skill workflow includes generating, running, and testing CLI code (e.g., execaCommand in references/node-cli.md and runner.invoke in references/python-cli.md).
Sanitization: The instructions do not specify validation or sanitization of user-provided strings before they are incorporated into generated code, templates, or help text.
[DYNAMIC_EXECUTION]: The reference guides (references/design-patterns.md) describe standard patterns for dynamic plugin loading from computed local paths (~/.mycli/plugins/) and environment variables (MYCLI_PLUGIN_PATH). While these are common extensibility patterns for CLI tools, they involve executing code from external or user-defined directories at runtime.
[SAFE]: The skill references standard, highly reputable libraries for CLI development across multiple languages, including commander, yargs, click, typer, and cobra.
[SAFE]: The documentation includes security best practices, such as recommending restrictive file permissions (600) for local credential storage files (references/design-patterns.md).

cli-developer