Plaintext-AI

Open-source AI research assistant for exploring academic papers, generating literature reviews, and extracting insights.

---

Quick Start

git clone https://github.com/l-forster/plaintext-ai.git
cd plaintext-ai
echo "OPENAI_API_KEY=sk-your-key-here" > frontend/.env
npm install && cd frontend && npm install && npm run dev

Open http://localhost:5173

That's it. One OpenAI API key required.

---

Features

• AI Research Agent - Ask questions about papers in natural language
• Paper Search - 200M+ papers from Semantic Scholar
• PDF Upload & Parsing - Extract metadata, claims, and references from PDFs
• Citation Network Visualization - Explore paper citation graphs
• Literature Review Generator - Auto-generate comprehensive reviews
• Claim Extraction - Extract key claims from research
• Contradiction Checker - Find contradictions in scientific text
• Visual Workflow Builder - Drag-and-drop research pipelines
• Citation Management - APA, MLA, Chicago, Harvard formats
• Data Analysis - CSV upload for AI insights
• Export - TXT, Word, Google Docs

---

Usage Examples

Research Agent

You: "Find recent papers on transformer architectures and summarize key contributions"

AI: *searches papers* → *analyzes* → *generates summary with citations*

Workflow Builder

Build custom pipelines:

[Source Finder] → [Claim Extractor] → [Contradiction Checker] → [Export DOC]

Run the workflow and get automated results.

Data Analysis

Upload a CSV file:

sales_data.csv → AI analyzes trends, outliers, correlations → Generate insights

PDF Upload

Upload academic PDFs:

paper.pdf → Extracts title, authors, abstract, claims, references → Search for related papers

Citation Network

Visualize citation relationships:

Enter paper ID → Builds citation graph → Explore connected papers interactively

---

SDK / API Usage

Using the API Client

import { apiRequest } from './lib/queryClient';

// Search papers
const papers = await apiRequest('POST', '/api/source-finder/search', {
  query: 'machine learning',
  model: 'nineveh'
});

// Generate literature review
const review = await apiRequest('POST', '/api/literature-review/generate', {
  reviewTopicScope: 'transformer architectures',
  reviewType: 'systematic',
  reviewDepthLength: 'comprehensive',
  reviewTone: 'academic',
  model: 'nineveh'
});

// Extract claims
const claims = await apiRequest('POST', '/api/claim-extractor/extract', {
  prompt: 'Large language models demonstrate emergent abilities...',
  model: 'nineveh'
});

// Check contradictions
const contradictions = await apiRequest('POST', '/api/contradiction-check/check', {
  text: 'Some scientific text to check...',
  modelId: 'nineveh'
});

// Format references
const formatted = await apiRequest('POST', '/api/reference-management/format', {
  referencesInput: 'Smith, J. (2023). Paper title...',
  citationStyle: 'apa',
  model: 'nineveh'
});

// Analyze CSV data
const formData = new FormData();
formData.append('file', csvFile);
const analysis = await fetch('/api/data-analysis/analyze', {
  method: 'POST',
  body: formData
});

Available API Endpoints

Endpoint	Method	Description
/api/scholar-ai-query	POST	AI research agent query
/api/source-finder/search	POST	Search academic papers
/api/literature-review/generate	POST	Generate literature review
/api/claim-extractor/extract	POST	Extract claims from text
/api/contradiction-check/check	POST	Check for contradictions
/api/reference-management/format	POST	Format citations
/api/data-analysis/analyze	POST	Analyze CSV data
/api/pdf/parse	POST	Parse PDF and extract metadata
/api/citations/:paperId/network	GET	Build citation network graph
/api/citations/:paperId/citations	GET	Get papers citing this paper
/api/citations/:paperId/references	GET	Get papers referenced by this paper
/api/export-tools/txt	POST	Export to TXT
/api/export-tools/doc	POST	Export to Word
/api/export-tools/google-doc	POST	Export to Google Docs
/semantic-scholar/search	GET	Proxy to Semantic Scholar

Request/Response Examples

Search Papers:

// Request
POST /api/source-finder/search
{
  "query": "neural networks",
  "model": "nineveh"
}

// Response
{
  "papers": [
    {
      "arxiv_id": "1234.5678",
      "title": "Deep Neural Networks",
      "summary": "Abstract text...",
      "authors": ["John Doe", "Jane Smith"],
      "published": "2023-01-15",
      "citations": 42,
      "url": "https://arxiv.org/abs/1234.5678"
    }
  ]
}

Generate Literature Review:

// Request
POST /api/literature-review/generate
{
  "reviewTopicScope": "attention mechanisms in transformers",
  "reviewType": "narrative",
  "reviewDepthLength": "moderate",
  "reviewTone": "academic",
  "yearFrom": 2020,
  "yearTo": 2024,
  "model": "nineveh"
}

// Response
{
  "review": "# Literature Review: Attention Mechanisms\n\n## Introduction\n...",
  "papers": [...],
  "citations": [...]
}

Extract Claims:

// Request
POST /api/claim-extractor/extract
{
  "prompt": "Recent studies show that large language models...",
  "model": "nineveh"
}

// Response
{
  "claims": [
    {
      "id": "claim_1",
      "text": "Large language models demonstrate emergent capabilities",
      "confidence": 0.95
    }
  ]
}

Using the Workflow Builder Component

import { WorkflowBuilder } from '@/components/workflow/WorkflowBuilder';

function MyApp() {
  return (
    <WorkflowBuilder
      onWorkflowComplete={(results) => {
        console.log('Workflow results:', results);
      }}
    />
  );
}

---

Development

Project Structure

plaintext-ai/
├── frontend/
│   ├── client/src/
│   │   ├── components/
│   │   │   ├── workflow/      # Workflow builder
│   │   │   └── ui/            # UI components
│   │   ├── pages/
│   │   │   ├── research.tsx   # Research agent page
│   │   │   └── workflow.tsx   # Workflow builder page
│   │   ├── lib/
│   │   │   ├── api.ts         # API client
│   │   │   └── queryClient.ts # React Query setup
│   │   └── types/             # TypeScript types
│   ├── server/
│   │   ├── routes/            # API endpoints
│   │   │   ├── aiAssistantRoutes.ts
│   │   │   ├── sourceFinder.ts
│   │   │   ├── literatureReview.ts
│   │   │   ├── claimExtractor.ts
│   │   │   ├── contradictionChecker.ts
│   │   │   ├── referenceManagement.ts
│   │   │   └── dataAnalysis.ts
│   │   └── services/          # Business logic
│   └── package.json
└── package.json

Tech Stack

Frontend:

• React 19 + TypeScript
• Vite
• TanStack Query (data fetching)
• Radix UI + Tailwind CSS
• React Flow (workflow visualization)
• Wouter (routing)

Backend:

• Express.js
• OpenAI API (GPT models)
• Semantic Scholar API

Development Commands

# Start dev server (both frontend and backend)
npm run dev


### Environment Variables

Create `frontend/.env`:

```bash
# Required
OPENAI_API_KEY=sk-your-openai-key

# Optional - for higher Semantic Scholar rate limits
SERVER_SEMANTIC_SCHOLAR_API_KEY=your-key

# Server config
NODE_ENV=development
PORT=3000

---

Available AI Models

Model ID	Name	Description
nineveh	Nineveh	Fast, cost-effective model for general queries
babylon	Babylon	Advanced model with reasoning capabilities

---

Tools Reference

Tool	Input	Output
Source Finder	Search query	List of papers
AI Literature Review	Topic, scope, type	Formatted review
Reference Management	Raw references	Formatted citations
Claim Extractor	Text	List of claims
Contradiction Checker	Text	Contradictions found
Data Analysis	CSV file	Insights & visualizations
PDF Upload	PDF file	Metadata, claims, references
Citation Network	Paper ID	Interactive graph visualization
Export TXT	Content	Text file
Export DOC	Content	Word document
Export Google Doc	Content	Google Docs link

---

Contributing

Pull requests welcome. Please ensure:

• Code follows existing style
• TypeScript types are properly defined
• API endpoints are documented

---

License

MIT License - see LICENSE file