forgelisted

Designs Claude Code harness artifacts. One triage ladder picks the right surface (skill, hook, path-scoped rule, CLAUDE.md or AGENTS.md entry, workflow script, subagent, MCP); the stance holds the bar every artifact must clear (a gradeable objective, a gap between the model's default and what the owner wants, and a nameable attention shift; the gap is a new idea or source worth operationalizing, a miscommunication to prevent, or an observed failure); per-surface mechanics live in references. Use for creating, reviewing, or refactoring any of these, for distilling articles or docs into one, for skills that under- or over-trigger, hooks that don't fire or fire too often, bloated CLAUDE.mds, and for deciding which surface a behavior belongs on. Triggers on any work on a skill, hook, rule, CLAUDE.md, AGENTS.md, or workflow orchestration script, including debugging triggering or firing semantics and packaging a /command or plugin.
claudialnathan/agent-kitchen · ★ 3 · AI & Automation · score 67

Install: claude install-skill claudialnathan/agent-kitchen

# forge  A skill transplants expertise. Its only measure is the work it makes the agent produce. That work is judged at the output, against the standard of the best practitioner alive in the domain — and, above that, against what this owner wanted it to achieve, because expert-grade in the wrong direction is still wrong. The bar is never generically good; it is exactly right for this owner. Everything else in this file (surfaces, triggering, token cost, lifecycle) is plumbing that gets that expertise in front of the agent at the right moment. A flawlessly-packaged artifact encoding mediocre expertise is worthless; a roughly-packaged one encoding genuine expertise is valuable. Optimize in that order, always. So the first question is never *what surface?* or *what will this cost?* It is **what output does this artifact exist to produce?** Every artifact carries an objective: a sentence or two naming the outcome its work is graded against, concrete enough that a stranger could judge an output with it ("layouts as elegant, responsive, and accessible as the page allows, in as few lines as it permits" grades; "helps with layouts" doesn't). The objective is the spec: the body serves it, the eval grades against it, the probes derive their pass criteria from it. And it comes from the owner before it comes from any source, because the model improves every release without the owner's help, while what the ow