PySpur - AI Agent Visual Development Platform

Project Overview

PySpur is a visual workflow platform for AI agents, enabling AI engineers to iterate and develop AI agents 10x faster. This is an open-source project backed by Y Combinator, designed to address the key pain points AI engineers face when building intelligent agents.

Core Problems Solved

AI engineers commonly face three major challenges when building AI agents:

1. Prompt Engineering Hell: Spending excessive time on prompt tuning and iterative trial and error.
2. Workflow Blind Spots: Lack of visualization of step interactions, leading to hidden failures and confusion.
3. End-to-End Testing Nightmare: Needing to stare at raw output and manually parse JSON.

Core Features

🔄 Workflow Management

• Visual Graphical Interface: Build AI workflows through drag-and-drop.
• Loop Support: Supports iterative tool calls with memory.
• Human-in-the-Loop: Persist workflows and support waiting for human approval.
• Breakpoint Debugging: Workflow pause points that require manual approval to continue execution.

📤 Multi-Modal Data Processing

• File Upload: Supports uploading files or pasting URLs to process documents.
• Multi-Modal Support: Handles various formats such as video, images, audio, text, and code.
• Structured Output: Provides a UI editor for JSON Schema.

🗃️ RAG System

• Complete RAG Process: Parses, chunks, embeds, and inserts data into vector databases.
• Vector Database Integration: Supports various vector databases.

🧰 Tool Integration

• Rich Tool Support: Integrates with Slack, Firecrawl.dev, Google Sheets, GitHub, etc.
• Extensibility: Add new nodes by creating a single Python file.

📊 Monitoring & Evaluation

• Automatic Tracking: Automatically captures the execution traces of deployed agents.
• Evaluation System: Evaluates agent performance on real-world datasets.
• One-Click Deployment: Publish as an API and integrate anywhere.

🎛️ Multi-Vendor Support

• 100+ Provider Support: Supports over 100 LLM providers, embedders, and vector databases.
• Python-Driven: Built on Python, easy to extend and customize.

Quick Start

Installation Requirements

• Python 3.11 or higher

Basic Installation Steps

1. Install PySpur

pip install pyspur

2. Initialize a New Project

pyspur init my-project
cd my-project

3. Start the Server

pyspur serve --sqlite

By default, this will start the PySpur application at http://localhost:6080 using a sqlite database. It is recommended to configure a postgres instance URL in the .env file for a more stable experience.

4. Configure Environment and API Keys (Optional)

• Application Interface Method: Navigate to the API Keys tab to add provider keys (OpenAI, Anthropic, etc.).
• Manual Method: Edit the .env file (recommended to configure postgres) and restart using pyspur serve.

Development Environment Setup

Recommended Method: Using Development Containers

It is recommended to use Cursor/VS Code with a development container (.devcontainer/devcontainer.json) to get:

• A consistent development environment with pre-configured tools and extensions.
• Optimized settings for Python and TypeScript development.
• Automatic hot reloading and port forwarding.

Steps:

1. Install Cursor/VS Code and the Dev Containers extension.
2. Clone and open the repository.
3. Click "Reopen in Container" when prompted.

Manual Setup Method

1. Clone the Repository

git clone https://github.com/PySpur-com/pyspur.git
cd pyspur

2. Start with docker-compose

docker compose -f docker-compose.dev.yml up --build -d

3. Custom Settings: Edit .env to configure the environment (e.g., PostgreSQL settings).

Note: Manual setup requires additional configuration and may not include all development container features.

Use Cases

PySpur is particularly suitable for the following scenarios:

• Complex AI workflows requiring visual debugging.
• Quality assurance processes requiring human supervision.
• Multi-modal data processing applications.
• RAG system construction and optimization.
• Large-scale intelligent agent application deployment.

Technical Architecture

• Frontend: Workflow editor based on a graphical interface.
• Backend: Python-driven execution engine.
• Database: Supports SQLite and PostgreSQL.
• Deployment: Supports containerized deployment and one-click API publishing.