MervinPraison/PraisonAI

PraisonAI 🦞 — Hire a 24/7 AI Workforce. Stop writing boilerplate and start shipping autonomous self-improving agents that research, plan, code, and execute tasks. Deployed in 5 lines of code with built-in memory, RAG, and support for 100+ LLMs.

📚 Quick Summary

PraisonAI is a Python‑first framework for building autonomous AI agents that can plan, act, and improve themselves. It ships a layered “Five‑Layer Agent Stack” (Prompt → Context → Harness → Loop → Graph) plus a “Managed Agents” layer that lets you run the heavy‑lifting (tool execution, even the whole agent) in remote sandboxes such as Docker, e2b, Modal, etc. The project includes:

Core SDK (praisonaiagents) for code‑first development.
CLI (praisonai) and optional UI/dashboard extensions.
JavaScript SDK.
Support for 100+ LLM providers.
Example agents for research, coding, content, data pipelines, support bots, and workflow automation.

🛠️ Core Components

Component	What it does	How you use it
Agent	A single LLM‑backed worker with memory, tools, and guardrails.	`Agent(role="Senior Data Analyst").start("Analyze trends")`
AgentFlow	Graph‑oriented orchestration of multiple agents (routing, parallel, repeat).	`AgentFlow(steps=[classifier, route(...), parallel([...])])`
ExecutionConfig	Controls loop limits, budget, and doom‑loop detection.	`execution=ExecutionConfig(max_iter=30, max_budget=0.5)`
Managed Runtime	Runs tools or the whole agent in a remote sandbox (Docker, e2b, Modal, etc.).	`tools_run_on="docker"` or `run_on="anthropic"`
MCP Protocol	Unified way to expose tools via stdio, HTTP, WebSocket, SSE.	`tools=[MCP("npx ...")]`
CLI	Manage sandboxes, list sessions, start agents from the terminal.	`praisonai managed ps`

🚀 Getting Started (Python)

pip install praisonaiagents   # core SDK
export OPENAI_API_KEY=…

from praisonaiagents import Agent
agent = Agent(instructions="You are a senior data analyst.")
agent.start("Analyze the top 3 tech trends of 2026 and format as markdown")

For remote execution:

agent = Agent(name="builder", instructions="You build things.", tools_run_on="docker")
agent.start("Write a script that prints primes, then run it")

📦 Ecosystem Extras

CLI (praisonai) – install with pip install praisonai for terminal control.
Claw Dashboard – connect agents to Telegram/Slack/Discord (pip install "praisonai[claw]").
Flow Visual Builder – drag‑and‑drop workflow editor (pip install "praisonai[flow]").
UI – chat‑style interface (pip install "praisonai[ui]").
JavaScript SDK – npm install praisonai for Node/Browser usage.

🎯 Typical Use Cases

Domain	Example
Research & Analysis	Auto‑gather papers, summarize, produce insight tables
Code Generation	Write, debug, refactor code across a repo
Content Creation	Multi‑agent teams produce blogs, docs, marketing copy
Data Pipelines	Pull from APIs/DBs, transform, and visualise results
Customer Support	24/7 bots on chat platforms with memory and knowledge bases
Workflow Automation	Chain agents to triage tickets, run tests, deploy builds

📚 Documentation & Resources

Full docs: docs.praison.ai
20 runnable examples in examples/python/managed-agents/
Provider‑specific snippets under examples/python/providers/
Visual flow can be expressed in YAML, no Python required.

🏷️ License & Community

MIT License – free for commercial and personal use.
Install stats badge shows total downloads.
Trending on Trendshift and highlighted on X (formerly Twitter).

🔗 Quick Links

GitHub: https://github.com/MervinPraison/PraisonAI
Download binaries: macOS / Windows / Linux badges in README
Installation script: curl -fsSL https://praison.ai/install.sh | bash

PraisonAI aims to give developers a complete stack for building, deploying, and managing autonomous AI agents without having to stitch together multiple libraries.