For skill authors: take a Claude Agent Skill from blank file to A+, tested, and packed.
The authoring loop for people who build Claude Agent Skills - the same toolchain Skill Me uses to hold its own catalog to a paid-quality bar. Reach for it when a skill needs to go from idea to shippable: draft a new one against the nine-part anatomy (procedure, elicitation, thresholds, worked artifact, deliverable, Do NOT, quality bar), audit an existing one to an absolute A+ with ship-ready rewrites, sharpen the description with the WHAT + WHEN + NOT trigger formula, prove the triggers fire with live subagent tests, and compose members into a pack with an install sequence and cross-links. End state is a skill (or pack) you can publish with confidence, not a to-do list.
Click to play with sound.
Arranged in the author's recommended order. Walk through them in sequence, or open any one on its own.
Writes a complete, ship-ready SKILL.md from a capability idea – a trigger-precise description plus procedure, input questions, thresholds, a worked example, the deliverable, a Do NOT list and a quality bar.
Grades whole skills and packs against an absolute A+ bar – precise triggers, no collisions with other skills, the full operating anatomy – and hands back ship-ready SKILL.md rewrites instead of a report.
Writes the description field that makes a skill fire reliably – WHAT it does, WHEN to invoke it with quoted user phrasing, and what it is NOT for.
Empirically tests a skill with subagent scenarios to verify it fires on the right prompts, stays silent on the wrong ones, and performs its job when loaded.
Designs the membership and structure of a skill pack – persona, member selection, install sequence, cross-links, and what to add or cut.