¿Puedo vibecodear Lex?

Build a personal replacement for Lex in an empty repository.
Use Next.js 15, TypeScript, Tailwind CSS, SQLite, Drizzle ORM, and the OpenAI Responses API; do not offer alternative stacks.
The core loop is: take a brief, gather user-supplied source material, generate and revise a structured long-form draft with comments and AI actions, and keep citations and version history attached to each section.
Make the first run work locally with one documented command.
Store all user data locally by default and make export straightforward.
Put secrets in .env, ship .env.example, and never commit credentials.
Build a brief form with audience, objective, tone, source URLs, and prohibited claims.
Store imported source text locally and chunk it for retrieval with SQLite FTS5.
Generate an outline first and require approval before drafting sections.
Attach source references to generated paragraphs and flag unsupported claims.
Provide rewrite controls for shorten, clarify, change tone, and add evidence.
Export clean Markdown plus a JSON research bundle.
Include clear empty, loading, success, and recoverable error states.
Add input validation, safe filenames, and graceful handling of unavailable APIs.
Write focused tests for the core transformation and one end-to-end happy path.
Create a README with setup, architecture, permissions, data location, and backup steps.
Do not add accounts, billing, telemetry, analytics, or a hosted control plane.
Deliberately leave out live search-engine rank data.
Deliberately leave out automatic publishing to third-party CMSs.
Deliberately leave out multi-user approvals and brand governance.
Finish by running the tests and listing the exact commands used.