Story Protocol Article Generator

This project generates articles based on Farcaster interactions and registers them with Story Protocol. It uses AI to analyze your Farcaster activity and create personalized articles.

Overview

The system:

1. Fetches your Farcaster casts and reactions
2. Processes them as "memories" (short-term and long-term)
3. Generates an article using Gaia's LLM
4. Registers the article with Story Protocol

Setup

1. Clone the repository

git clone [your-repo-url]
cd storyprompts§

2. Install dependencies

npm install

3. Configure environment variables:
   • Copy .env.example to .env
   • Fill in your credentials:

     # Neynar API Configuration
     NEYNAR_API_KEY=your_neynar_api_key
     
     # Story Protocol Wallet Configuration
     WALLET_ADDRESS=your_wallet_address
     WALLET_PRIVATE_KEY=your_wallet_private_key
     
     # Pinata Configuration
     PINATA_JWT=your_pinata_jwt
     
     # RPC Configuration
     RPC_PROVIDER_URL=your_rpc_provider_url

Usage

Generate and Register an Article

Run docker for chromadb database.

docker run -p 8000:8000 chromadb/chroma 

npm run start "Your Article Title"

Run Tests

# Test article generation only
npm run test:generatearticle

# Test memory processing
npm run test:memory

# Test full flow (generation + registration)
npm run test:main "Test Article Title"

API Endpoints Used

• Neynar API: https://api.neynar.com/v2
• Gaia Node: https://llama8b.gaia.domains/v1

Current Limitations

1. Farcaster Data:

   • Limited to 100 reactions
   • Hardcoded to user ID 12021
   • Only processes parent URLs (not channel IDs)

2. Article Generation:

   • Token limit: 81,920 tokens
   • Currently limited to:
     • 3 long-term memory posts
     • 5 core interest casts
     • 3 related topic casts
     • 2 outer space casts

Project Structure

src/
├── services/
│   ├── article.ts     # Article generation logic
│   ├── feed.ts        # Farcaster feed fetching
│   ├── memory.ts      # Memory processing
│   └── reactions.ts   # Farcaster reactions handling
├── test/
│   ├── test-article.ts
│   ├── test-main.ts
│   └── test-memory.ts
└── utils/            # Utility functions

Known Issues

1. Channel ID handling needs improvement
2. URL metadata is not currently processed

Future Improvements

• Improve error handling for API responses

data fetching and processing

• Add time range filtering (30 days limit)
• dynamic limit for number of casts to fetch: First time fetch all, second time only new casts
• Improve channel handling (from parent_channel field to channel-id field)
• Add URL metadata processing
• Add image description capabilities
• reaction fetching is still limited to first 100 reactions.

article generation

• Add more memories
• Add more casts
• Include source URLs of casts in article sections

Technical Details

• Uses Gaia's LLM for article generation
• Implements Story Protocol for IP registration
• Uses Pinata for IPFS storage
• Processes Farcaster data through Neynar API

License

ISC

Project tasks

1. Cast 0: Create a webhook that listens to mentions of the bot
2. Cast 1: Default bot reply 1: we'll start with a topic. My human team member strongly believes that human's unconsciousness is embedded in their writing. I'll analyze 100 of your casts and reply with a couple of themes for you to pick from. If you have time (and not the attention span of a gold fish), wait. If not, no hard feelings, I'll tag you.
3. Cast 2 (reply to cast 1): Pick a theme for the story: list of themes (limit to 5)