Skill Evals

Evaluate this repo's skills with PluginEval (pulled on demand via uvx from wshobson/agents — nothing is vendored) and drop a Markdown report into each skill's own folder as EVALS.md. The work fans out across one subagent per skill so reports are produced in parallel and each subagent's verbose uvx/LLM output stays out of the main context.

The upstream docs are bundled so you can look them up without network access:

• references/plugin-eval.md — the eval framework (commands, depths, layers, dimensions, badges).
• references/agent-skills-how-skills-work.md — what a good skill looks like (progressive disclosure, triggering, spec rules).
• references/fix-playbook.md — translating a weak dimension into a concrete SKILL.md edit (used by --fix).

Arguments