{
  "uid": "cogitave.learn.reuse-first-engineering.introduction",
  "kind": "moduleUnit",
  "href": "/modules/reuse-first-engineering/introduction/",
  "title": "Introduction",
  "summary": "",
  "type": null,
  "products": [],
  "roles": [],
  "levels": [],
  "subjects": [],
  "headings": [
    "What you will get from this module"
  ],
  "source": "Agents write most of Cogitave's code - `yuva`/`namzu` and external coding agents\nalike. Left to itself, that is a liability. A large language model generates by\nstatistical prediction: the same problem, asked twice, yields two similar-but-not-\nidentical solutions. Most AI review tooling only sees the diff, so a freshly\ngenerated helper that already exists elsewhere is invisible to it. Unchecked,\n**more AI means more divergence** - a fourth Button, a near-duplicate date\nformatter, a service that reinvents a primitive three repos already have.\n\nCogitave is agent-first and **full-scratch over one canonical model**\n([Cogitave Core](../../../../core/docs/architecture.md)); divergence is not a style\nnit here, it undercuts the very thing that makes the estate harmonic. The\ncounter-force is deliberate: **retrieval before generation**. That operating\nmodel is the\n[AI-Native Development standard](../../../../standards/docs/standards/ai-native-development.md)\n(decision: [ADR-0013](../../../../standards/docs/decisions/0013-reuse-first-convergent-ai-development.md)).\n\n> [!IMPORTANT]\n> Before authoring any UI or code, every agent and every human MUST query the\n> registry and the design system for an existing fit. **Authoring something new\n> is the exception**, requires a recorded \"no fit\" justification, and is\n> contributed back to the shared library - never kept as a local fork.\n\nThe same discipline applies one level up, to whole engines and tools, not just\ncomponents: [ADR-0003](../../../../standards/docs/decisions/0003-build-from-scratch-reference-not-dependency.md)\ngoverns when \"build it ourselves\" is actually earned versus when it is blind\nreinvention of a problem the industry already solved well.\n\n## What you will get from this module\n\n- **Why reuse-first, and what enforces it** - the discover-before-generate hard\n  rule, the tokens-to-primitives-to-golden-components single source of truth, the\n  MCP-native registry, and the anti-duplication gates that keep a needless\n  re-author out of `main`.\n- **Build from scratch as the exception** - ADR-0003's port-best-in-class,\n  rebuild-native stance: references are studied and ported, not reinvented\n  blindly, and a from-scratch rebuild has to be materially earned.\n\nBy the end, you will be able to state the hard rule, name what enforces it, and\njustify a from-scratch decision the way the estate's own ADRs require - as a\nreasoned exception, never a default.\n",
  "partOf": "cogitave.learn.reuse-first-engineering",
  "durationInMinutes": 4,
  "quiz": null
}