skill-based-architecture

name	skill-based-architecture
description	This skill should be used when the user asks to "organize the project rules", "clean up scattered documentation", "把规则迁移到 skills 目录", "优化 skill 路由", "提高 description 命中率", or "减少薄壳重复维护". Activate when a SKILL.md is too large, rules are duplicated across agent entry files, task routing or trigger_examples miss natural user language, or templates / thin shells / validation scripts need drift-resistant maintenance.
primary	true

skill-based-architecture (Cursor Registration Entry)

This is the Cursor discovery entry. Formal skill content lives at the repo root — read SKILL.md immediately, then follow its design principles and workflow routing.

Quick Routing (survives context truncation)

Task routes live in references/self-hosting-routing.yaml.

For every new task:

1. Read SKILL.md.
2. Read references/self-hosting-routing.yaml.
3. Match exactly one route by labels, trigger_examples, and task intent; if none matches, use other.
4. Follow only that route's workflow; the route does not preload project knowledge.
5. Let the workflow inspect the smallest evidence that can decide the next action, then pull later knowledge only when that unresolved decision requires it.

Why this file exists

Cursor's agent_skill mechanism only scans .cursor/skills/. Without this registration entry, the formal skill at repo root is invisible to Cursor.

The description above must stay identical to the root SKILL.md description field. Drift between the two = Cursor uses one activation rule while other harnesses use another. templates/skill/scripts/smoke-test.sh automatically verifies they match.