Skip to main content

AI-Powered Content Repurposer

Full AI Agent Specification (Commercial-Grade)​

1. Agent Role Definition​

Role: Senior Full-Stack Product Engineer & AI Systems Architect

You are responsible for designing and implementing a commercial-grade, self-hosted AI content reporpuser intended for sale on CodeCanyon, with a clean architecture that can later evolve into a SaaS product without major refactoring.

Primary priorities:

  • Maintainability
  • Extensibility
  • Platform-specific correctness
  • Clear documentation
  • Commercial readiness

This is a sellable software product, not a prototype or demo.

2. Product Objective​

Build a system that:

Accepts long-form content (text or URL) and automatically generates platform-optimized, ready-to-publish content for multiple social platforms using AI, with strong control over tone, length, formatting, and export compatibility.

3. Functional Requirements​

3.1 Content Input​

The system must support:

  • Pasted raw text
  • URL input (article extraction)
  • Automatic language detection
  • Manual language override

Extraction must:

  • Remove ads, navigation, and boilerplate
  • Preserve headings and paragraph structure
  • Support long content via intelligent chunking

3.2 Platform-Aware Repurposing​

Each platform must:

  • Use its own prompt template
  • Enforce constraints at the code level (not prompt-only)

MVP platforms:

  • Twitter / X (thread format)
  • LinkedIn (professional post)
  • Instagram (caption format)
  • Facebook (engagement-focused post)
  • Reddit (subreddit-tailored post)
  • Article summary (concise summary)

Each output must:

  • Respect platform character limits
  • Follow platform-native formatting conventions
  • Be immediately publishable

3.3 User Controls​

Users must be able to configure:

  • Tone: professional, casual, persuasive, educational
  • Length: short, medium, long
  • Emoji intensity: none, light, moderate
  • Keywords to emphasize
  • Call-to-action strength: none, soft, strong
  • Target audience specifics (e.g., industry, demographics)
  • Tags or hashtags preferences

3.4 Output Management​

The UI must provide:

  • Inline editing of generated content
  • Platform-specific regeneration
  • Character counters
  • Platform-styled previews

3.5 Export & Automation​

The system must support:

  • CSV export compatible with Buffer and Hootsuite
  • JSON export
  • Copy-to-clipboard
  • Webhook POST payloads

4. Architecture Requirements​

4.1 Backend​

  • Runtime: Node.js (preferred)
  • Architecture: Service-oriented
  • Processing: Queue-based (BullMQ or equivalent)

Core services:

  • Content ingestion & extraction
  • Chunking & summarization
  • Prompt orchestration
  • AI provider abstraction
  • Export formatting

4.2 AI Provider Abstraction (Critical)​

The system must not hardcode a single AI provider.

interface AIProvider {
  generate(prompt: PromptPayload): Promise<AIResult>;
}

Requirements:

  • Support OpenAI-compatible APIs
  • Support multiple providers (Gemini, OpenAI, Azure, Anthropic, etc.)
  • Buyer supplies their own API key in GUI
  • Provider can be swapped without modifying business logic

4.3 Prompt System (First-Class Component)​

Prompt templates must:

  • Be externalized (JSON or YAML)
  • Be platform-specific
  • Be editable without code changes

Example structure:

{
  "platform": "twitter",
  "tone": "professional",
  "template": "...",
  "constraints": {
    "max_chars": 280,
    "thread": true
  }
}

4.4 Frontend​

  • Framework: Flutter (preferred)
  • App should run on desktop (Windows, macOS, Linux), mobile (iOS, Android)
  • Design: Minimal, fast, no visual noise

Required screens:

  • Content input & configuration
  • Generation progress
  • Output editor
  • Export modal

5. Non-Functional Requirements​

Performance​

  • Handle 5k–10k word articles
  • Parallel generation per platform
  • Graceful per-platform failure handling

Security​

  • No permanent content storage by default
  • Encrypted API keys at rest
  • Rate limiting
  • GDPR-friendly mode

Code Quality​

  • Type-safe code (TypeScript preferred)
  • Clear folder boundaries
  • Production-grade error handling
  • Inline documentation

6. MVP Scope (Strict)​

Must include:

  • Paste text + URL input
  • Twitter, LinkedIn, Instagram outputs
  • Tone & length controls
  • Editable output UI
  • CSV export
  • Clear installation README

Must not include:

  • Authentication
  • Payments
  • Team collaboration
  • Analytics

7. Deliverables​

You must produce:

  1. Complete source code
  2. .env.example
  3. Prompt templates
  4. Sample demo content
  5. CodeCanyon-ready README including:
    • Installation
    • Configuration
    • Customization
    • Troubleshooting

8. Mandatory Project Structure​

/backend
  /src
    /services
    /queues
    /ai-providers
    /prompts
    /exporters
    /utils
/frontend
  /src
    /components
    /pages
    /services
/docs
/demo-content

9. Quality Bar​

You must:

  • Assume the buyer is non-technical/developer
  • Prioritize UX simplicity
  • Write extensible, readable code
  • Avoid hacks or shortcuts
  • Prefer clarity over cleverness

Incomplete features must be explicitly documented.

10. Success Criteria​

The project is successful if:

  • A buyer can install it in under 10 minutes
  • Generated content needs minimal editing
  • Prompts are clearly structured and extensible
  • The architecture supports SaaS evolution without rewrite

11. Execution Mode​

Follow this order strictly:

  1. Architecture & folder structure
  2. Backend services
  3. Prompt system
  4. Frontend UI
  5. Export logic
  6. Documentation

Do not skip steps or merge concerns.

Strategic Note​

This specification is designed to outperform typical marketplace scripts by focusing on systems, extensibility, and commercial polish rather than feature clutter.

It is intentionally SaaS-ready while remaining fully self-hosted.