harshaneel/humanize ? reverse-engineered prompt

Build me a small GitHub project for two reusable AI writing skills, one called humanize and one called ai check.

I want the humanize skill to rewrite text so it sounds more natural, relatable, and less robotic, and I want the ai check skill to read a passage and flag signs that it may be AI written, with a clear verdict and confidence. Keep both skills static, with no model calls or runtime dependencies, just well written instruction files and supporting docs.

Please include a simple install script so the skills can be copied or linked into common agent folders, plus a README that explains what each skill does, how to install it, and how to use it in plain language. If you need to look up current setup details for any of the supported agents, do that. Also add basic tests or checks if they make sense for validating the skill files and install flow.