PostFuel

PostFuel is a content machine created as a Next.js 14 App Router project for generating and reviewing AI-assisted content across four formats:

• Blog drafts
• Code walkthroughs
• Article or PDF summaries
• Social posts

The app is designed around one shared idea: every draft should keep evidence attached so an editor can review claims before sharing or publishing.

What is implemented

Studio workflow

• / is the main studio page.
• Users choose a format, enter a topic, add notes or a URL, and optionally upload a PDF for summaries.
• Submitting the form calls one of the agent API routes and creates a saved draft.

Review workflow

• /drafts/[id] is the draft review page.
• The draft can be edited inline.
• Evidence items can be toggled as verified.
• Drafts can be copied as Markdown.
• Share helpers are included for email, X, and LinkedIn.

Agent routes

The app currently exposes these route handlers:

• POST /api/agents/blog
• POST /api/agents/code
• POST /api/agents/summary
• POST /api/agents/social
• GET /api/drafts
• GET /api/drafts/[id]
• PATCH /api/drafts/[id]
• POST /api/uploads
• GET /api/uploads/[filename]

Tech stack

• Next.js 14
• TypeScript
• Tailwind CSS
• Minimal shadcn-style UI primitives in src/components/ui
• OpenAI SDK for model-backed generation when configured
• react-markdown + rehype-highlight for draft preview
• jsdom + @mozilla/readability for URL/article extraction
• Optional Supabase persistence
• Optional Vercel Blob uploads
• Vitest for core generation tests

Project structure

src/
  app/
    page.tsx                  # Main studio UI
    drafts/[id]/page.tsx      # Review/edit/share screen
    api/
      agents/*/route.ts       # Four content agent endpoints
      drafts/route.ts         # List drafts
      drafts/[id]/route.ts    # Read/update a draft
      uploads/route.ts        # Upload PDFs
      uploads/[filename]/route.ts
  components/
    studio-form.tsx
    review-client.tsx
    recent-drafts.tsx
    markdown-renderer.tsx
    ui/
  lib/
    agents.ts                 # Agent orchestration and fallbacks
    draft-store.ts            # Supabase or local JSON persistence
    summary.ts                # URL extraction and chunking helpers
    uploads.ts                # Vercel Blob or local file storage
    research.ts               # Suggested web sources
    types.ts
data/
  drafts.json                 # Local fallback persistence

How the app behaves today

Persistence

The app uses two storage modes:

1. Supabase if both of these are set:
   • NEXT_PUBLIC_SUPABASE_URL
   • SUPABASE_SERVICE_ROLE_KEY
2. Local JSON fallback if Supabase is not configured:
   • Drafts are stored in data/drafts.json

Uploads

PDF uploads also support two modes:

1. Vercel Blob if BLOB_READ_WRITE_TOKEN is set
2. Local file fallback in data/uploads/ if Blob is not configured

AI generation

If OPENAI_API_KEY is set, the app uses OpenAI for draft generation.

If it is not set, the app falls back to deterministic template-based output so the UI and workflow still function locally.

Summary behavior

Summary generation supports:

• URL summarization through article extraction
• Text-based fallback summarization from notes
• Optional delegation to an external Python service for PDF-heavy workflows

If the Python service is not configured, the app still works, but uploaded PDFs do not go through a full PyMuPDF pipeline.

Local development

Install dependencies:

npm install

Start the app:

npm run dev

Open:

http://localhost:3000

Verification commands

These all pass in the current implementation:

npm run test
npm run lint
npm run build