lidr-skill-creator

Build, evaluate, and iteratively improve Claude Code skills through test-driven development and benchmarking. Domain-agnostic — works for any type of skill regardless of technology or industry context. Use for new skill creation, skill quality improvement, description optimization, and trigger testing. Essential when a team needs a new specialized workflow automated as a reusable skill. Always use when building a skill from scratch, always use when improving an existing skill's triggering or output quality. Do NOT use for one-off tasks that don't warrant reusable automation, or for command/hook/MCP/rule authoring (use lidr-agents-architecture). Triggers on "build a new skill", "create skill from scratch", "test skill performance", "benchmark skill", "evaluate skill", "improve skill description", "optimize triggering", "run skill evals", "measure skill variance". Output in English. Audience: Tech Lead (creates skills), Developer (uses and refines skills).

By LIDR-IT schedule Updated 6/12/2026