Production Ready

Complete AI SaaS Toolkit

Everything you need to build, secure, and scale your AI application. Battle-tested tools with comprehensive test coverage.

🤖

LangChain Agent Orchestration

Fully Tested

Multi-step reasoning with the ReAct pattern. Automatically selects and chains tools to solve complex queries.

✓6 integrated AI tools
✓Streaming responses with intermediate steps
✓Context-aware conversation history
✓Max 10 iterations with automatic loop prevention
✓POST /api/agent & /api/agent/stream endpoints

🌐

Web Tools Suite

Tested & Working ✓

Powerful web data extraction with DuckDuckGo search and JavaScript-rendering web scraper.

✓DuckDuckGo search (3-50 results)
✓News search & region-specific results
✓Playwright + Chromium for JS rendering
✓HTML → Markdown/Text conversion
✓Smart content filtering & metadata extraction

📄

Document Processing

Production Ready

Extract text, tables, and metadata from any document format with OCR support.

✓PDF: Text + tables (pdfplumber, max 1000 pages)
✓Office: DOCX paragraphs + tables
✓Excel: Multi-sheet support (XLSX)
✓PowerPoint: Slide-by-slide extraction (PPTX)
✓OCR support via pytesseract

💻

Sandboxed Code Interpreter

Fully Tested

Execute Python code safely in a restricted environment for data analysis and visualizations.

✓RestrictedPython environment
✓pandas, numpy, matplotlib available
✓30-second timeout protection
✓No file system or network access
✓Base64-encoded plot outputs

👁️

GPT-4 Vision Integration

Production Ready

Analyze images with state-of-the-art vision AI for OCR, chart analysis, and image comparison.

✓Image description & analysis
✓OCR from screenshots
✓Chart & graph interpretation
✓Multi-image comparison
✓Max 20MB per image

🛡️

Security Guardrails

All Tests Passing ✓

Protect your users and your business with PII detection, toxicity filtering, and prompt injection defense.

✓PII Detection: Email, phone, SSN, credit cards, API keys
✓Toxicity filtering with risk scores (0-1)
✓Prompt injection detection with confidence scores
✓Automatic redaction for sensitive data
✓Risk level classification (low/medium/high)

🔐

Role-Based Access Control (RBAC)

Production Ready

Fine-grained permission system for controlling access to AI models and features.

✓Model-specific permissions (GPT-4, Claude, Gemini)
✓Role hierarchy: Admin, Developer, User
✓API usage tracking & analytics
✓Failed attempt logging
✓Auto-role assignment for OAuth users

🔑

Multi-Provider Authentication

Fully Implemented

Flexible authentication with email/password, Google OAuth, and GitHub OAuth.

✓NextAuth.js integration
✓Google & GitHub OAuth
✓Email/password with bcrypt
✓JWT session management
✓Automatic permission loading

See It In Action

Research Agent

Autonomous multi-step research with web search and scraping

const response = await fetch('/api/agent', {
  method: 'POST',
  body: JSON.stringify({
    query: "Research quantum computing and summarize",
    system_prompt: "You are a technical researcher"
  })
});

// Agent automatically:
// 1. Searches web (DuckDuckGo)
// 2. Scrapes article content
// 3. Synthesizes summary

Data Analysis

Sandboxed Python execution with pandas, numpy, and matplotlib

const response = await fetch('/api/tools/code/execute', {
  method: 'POST',
  body: JSON.stringify({
    code: `
import pandas as pd
df = pd.DataFrame(sales_data)
print(df.describe())
df.plot(kind='bar')
plt.savefig('chart.png')
`,
    inputs: { sales_data: [...] }
  })
});

PII Protection

Detect and redact sensitive personal information

const response = await fetch('/api/guardrails/pii/redact', {
  method: 'POST',
  body: JSON.stringify({
    text: "Email me at john@example.com or call 555-1234"
  })
});

// Response:
{
  "redacted_text": "Email me at ████████████████ or call ████████",
  "changes": [
    { "type": "email", "original": "john@example.com" },
    { "type": "phone", "original": "555-1234" }
  ],
  "redactions_count": 2
}

Document Processing

Extract text and tables from PDF, DOCX, XLSX, PPTX

const formData = new FormData();
formData.append('file', pdfFile);

const response = await fetch('/api/tools/document/process', {
  method: 'POST',
  body: formData
});

// Returns: text, pages[], tables[], metadata

12 Production API Endpoints

POST

/api/agent

Run agent with tools

POST

/api/agent/stream

Streaming agent (SSE)

POST

/api/tools/web/search

DuckDuckGo search

POST

/api/tools/web/scrape

Scrape any URL

POST

/api/tools/code/execute

Run Python code

POST

/api/tools/vision/analyze

Analyze images

POST

/api/tools/documents/pdf

Extract PDF content

POST

/api/tools/documents/office

Process DOCX/XLSX

POST

/api/guardrails/pii

Detect & redact PII

POST

/api/guardrails/toxicity

Filter toxic content

POST

/api/guardrails/injection

Detect injections

POST

/api/v1/chat/completions

OpenAI-compatible chat

Ready to Build?

Get started with Docker in under 2 minutes:

# Clone the repository
git clone https://github.com/your-org/chimerai.git
cd chimerai

# Start all services with Docker Compose
docker-compose up -d

# Access your AI SaaS
- Frontend: http://localhost:3000
- API Docs: http://localhost:8001/docs
- Agent playground: /api/agent/stream

📚

Full Documentation

Comprehensive guides in /docs

✅

80+ Tests

Production-ready test coverage

🔧

Cursor & Copilot Ready

Optimized for AI-assisted development

Get Early Access

Follow on X Follow on LinkedIn