# PraisonAI

> PraisonAI is an open-source multi-agent AI framework that lets you build, deploy, and orchestrate autonomous AI agents with memory, RAG, MCP support, and 100+ LLM integrations in just 5 lines of code.

PraisonAI is an open-source Python and JavaScript framework for building autonomous, self-improving AI agents that can research, plan, code, and execute tasks across your applications. It supports single and multi-agent workflows, ships with built-in memory and RAG, and integrates with 100+ LLMs including OpenAI, Anthropic, Gemini, and local models via Ollama. The ecosystem spans a core SDK, CLI, visual flow builder, dashboard UI, and MCP protocol support — all deployable in 5 lines of code.

- **Multi-Agent Orchestration** — *Create sequential, parallel, or routed agent pipelines using `Agents([agent1, agent2])` or YAML config files.*
- **MCP Protocol Support** — *Connect agents to any MCP server via stdio, HTTP, WebSocket, or SSE transports using `tools=MCP(...)`.*
- **Built-in Memory & RAG** — *Enable short/long-term memory with `memory=True` and retrieval-augmented generation with zero extra dependencies.*
- **100+ LLM Providers** — *Switch between OpenAI, Anthropic, Gemini, DeepSeek, Mistral, Groq, Ollama, AWS Bedrock, Azure, and 20+ more.*
- **Claw Dashboard** — *Deploy agents to Telegram, Discord, Slack, and WhatsApp from a single web UI via `pip install "praisonai[claw]"`.*
- **Flow Visual Builder** — *Drag-and-drop multi-agent workflow creation powered by Langflow via `pip install "praisonai[flow]"`.*
- **Planning & Deep Research** — *Enable autonomous multi-step research and planning with `planning=True` and `--deep-research` CLI flags.*
- **Guardrails & Human Approval** — *Add input/output validation and human-in-the-loop checkpoints to any agent pipeline.*
- **Workflow Patterns** — *Use route, parallel, loop, and repeat patterns for complex agentic workflows.*
- **Telemetry & Tracing** — *Monitor agents with OpenTelemetry and Langfuse tracing out of the box.*
- **CLI Quick Start** — *Install with `pip install praisonai` or `curl -fsSL https://praison.ai/install.sh | bash` and run agents from the terminal.*
- **JavaScript SDK** — *Use `npm install praisonai` for Node.js-based agent development.*
- **Persistence** — *Persist agent sessions and conversations to PostgreSQL, MySQL, SQLite, MongoDB, Redis, and 20+ databases.*
- **Sandbox Execution** — *Run agent-generated code in isolated sandboxes for safe autonomous operation.*

## Features
- Single and multi-agent orchestration
- MCP protocol support (stdio, HTTP, WebSocket, SSE)
- Built-in memory (short and long-term)
- RAG and quality-based RAG
- 100+ LLM provider integrations
- Claw Dashboard for Telegram/Discord/Slack/WhatsApp
- Flow visual drag-and-drop workflow builder
- Deep research and planning mode
- Guardrails and human approval
- Workflow patterns: route, parallel, loop, repeat
- OpenTelemetry and Langfuse tracing
- CLI with interactive TUI
- JavaScript SDK
- Database persistence (PostgreSQL, MySQL, SQLite, MongoDB, Redis)
- Sandbox code execution
- Agent handoffs
- Self-reflection and reasoning agents
- Prompt caching
- Model router for cost optimization
- Shadow Git checkpoints and auto-rollback
- Context compaction
- Doom loop detection
- Graph memory (Neo4j-style)
- Background tasks
- Policy engine
- YAML-based no-code agent configuration
- 100+ built-in tools
- Multimodal agents (image, video, camera)
- Scheduling (cron-based agents)
- A2A agent-to-agent protocol

## Integrations
OpenAI, Anthropic, Google Gemini, DeepSeek, Azure OpenAI, Ollama, Groq, Mistral, Cerebras, Cohere, OpenRouter, Perplexity, Fireworks, AWS Bedrock, xAI Grok, Vertex AI, HuggingFace, Together AI, Databricks, Replicate, Cloudflare, AI21, SageMaker, vLLM, Moonshot, Langfuse, LangChain, Langflow, Telegram, Discord, Slack, WhatsApp, Tavily, Firecrawl, Crawl4AI, Mem0, Neo4j, PostgreSQL, MySQL, SQLite, MongoDB, Redis

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, DEVELOPER_SDK, CLI

## Pricing
Open Source

## Version
v4.6.37

## Links
- Website: https://docs.praison.ai
- Documentation: https://docs.praison.ai
- Repository: https://github.com/MervinPraison/PraisonAI
- EveryDev.ai: https://www.everydev.ai/tools/praisonai