building-commandslisted

# Building Commands <!-- uses: - skill: add-final-report - command: /add-framework--build - skill: building-commands/references/agent-dispatch.md - mention: add-framework-development - mention: add-review-discipline - mention: @prompt-review-agent --> <!-- `add.md` and `add.ux` below are PRODUCT commands and are named in prose on purpose. They are deliberately NOT declared: `uses:` targets resolve inside the declaring artefact's own layer (scripts/build.js), so `- command: /add.md` from here would resolve to `internal/command/add.md`, which does not exist, and the dangling gate would fail the build. The prose sniff skips cross-layer names. The same applies to the product `add-final-report`. `- skill: add-final-report` above resolves to the INTERNAL one, which is correct — the product sibling carries the same name and cannot be addressed from here at all. --> ## Overview **Commands fail not from unclear logic but from unclear PRESSURE POINTS.** Agents skip steps when gates are implicit, execute wrong order when sequence isn't mandatory, and rationalize when checklists are vague timelines instead of checkboxes. **Core principles:** 1. **Imperative > Informative** — Commands are ORDERS, not documentation 2. **Tool-specific prohibitions** — "DO NOT USE Grep" instead of "STOP" 3. **Gates block** — Execution path impossible to get wrong 4. **Checklists verify** — Checkboxes, not timelines --- ## When to Use **Symptoms you need this skill:** - Commands have optional "recomm