{"id":20961464,"url":"https://github.com/superagentxai/superagentx","last_synced_at":"2026-02-27T10:31:18.643Z","repository":{"id":259171184,"uuid":"876544822","full_name":"superagentxai/superagentx","owner":"superagentxai","description":"Move from idea to production in hours with policy-driven autonomous AI agents. Unified Control Plane: Centralised tools, MCPs, models, data, and policies with consistent observability and governance.","archived":false,"fork":false,"pushed_at":"2026-02-24T07:49:48.000Z","size":8381,"stargazers_count":182,"open_issues_count":1,"forks_count":40,"subscribers_count":10,"default_branch":"master","last_synced_at":"2026-02-24T13:51:06.287Z","etag":null,"topics":["agentic-ai","agentic-framework","agents","aigovernance","aisecurity","autonomous-agents","azure-openai","bedrock","claude-ai","deepseek","gemini","llms","memory","ollama","openai","orchestration","rag-pipeline"],"latest_commit_sha":null,"homepage":"https://superagentx.ai","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/superagentxai.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2024-10-22T06:40:24.000Z","updated_at":"2026-02-17T13:47:39.000Z","dependencies_parsed_at":"2024-11-10T04:18:05.912Z","dependency_job_id":"84cb49f2-5e5c-4e91-87e7-ae12e17d2dc1","html_url":"https://github.com/superagentxai/superagentx","commit_stats":null,"previous_names":["superagentxai/superagentx"],"tags_count":41,"template":false,"template_full_name":null,"purl":"pkg:github/superagentxai/superagentx","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/superagentxai%2Fsuperagentx","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/superagentxai%2Fsuperagentx/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/superagentxai%2Fsuperagentx/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/superagentxai%2Fsuperagentx/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/superagentxai","download_url":"https://codeload.github.com/superagentxai/superagentx/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/superagentxai%2Fsuperagentx/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29891408,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-27T09:48:51.284Z","status":"ssl_error","status_checked_at":"2026-02-27T09:48:43.992Z","response_time":57,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic-ai","agentic-framework","agents","aigovernance","aisecurity","autonomous-agents","azure-openai","bedrock","claude-ai","deepseek","gemini","llms","memory","ollama","openai","orchestration","rag-pipeline"],"created_at":"2024-11-19T02:13:49.549Z","updated_at":"2026-02-27T10:31:18.604Z","avatar_url":"https://github.com/superagentxai.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n\u003cimg src=\"https://github.com/superagentxai/superagentX/blob/master/docs/images/fulllogo_transparent.png\" width=\"350\"\u003e\n\n\n\u003cbr/\u003e\n\n\n**SuperAgentX**: is an open-source, modular agentic AI framework that enables AI agents to plan, act, and execute real-world workflows—with built-in human approval, governance, and auditability.\nUnlike traditional chatbots, SuperAgentX is designed for action, not just conversation.\n\n\n\n\u003cbr/\u003e\n\n[![Python 3.12+](https://img.shields.io/badge/python-3.12+-blue.svg)](https://www.python.org/downloads/release/python-31210/)\n[![GitHub Repo stars](https://img.shields.io/github/stars/superagentxai/superagentX)](https://github.com/superagentxai/superagentX)\n[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://github.com/superagentxai/superagentX/blob/master/LICENSE)\n\n\u003c/div\u003e\n\n## ✨ Why SuperAgentX?\n\nSuperAgentX enables AI agents to:\n- Execute multi-step workflows\n- Interact with browsers, APIs, databases, tools \u0026 MCPs\n- Pause for **human approval** before sensitive actions\n- Persist execution state, memory, and audit logs\n\nAll while keeping humans firmly in control.\n\n# Quick start\n\n```shell\npip install superagentx\n```\n\n## 🧠 Core Capabilities\n\n### 🔹 Massive Model \u0026 Tool Support\n- ✅ **100+ LLMs supported** (OpenAI, Azure OpenAI, Gemini, Claude, Bedrock, OSS models)\n- ✅ **10,000+ MCP (Model Context Protocol) tools supported**\n- ✅ **Browser Agents** using real browser automation (Playwright)\n\n---\n\n### 🔹 Agentic AI (Beyond Chatbots)\nAgents can:\n- Understand goals\n- Plan execution steps\n- Call tools dynamically\n- Run sequential or parallel workflows\n- Retry, reflect, and recover\n\n---\n\n### 🔹 Human-in-the-Loop Governance\nA built-in **Human Approval Governance Agent**:\n- Pauses sensitive actions\n- Requests explicit approval\n- Resumes or aborts execution\n- Persists decisions for audit\n\n➡️ AI **cannot act blindly**.\n\n---\n\n## 🗄️ Persistent Data Store \u0026 Memory\n\n### Supported Data Stores\n- 🗃 **SQLite** – lightweight, local workflows\n- 🗄 **PostgreSQL** – production-grade, multi-tenant systems\n\n### Stored Data\n- Workflow state\n- Agent decisions\n- Human approvals\n- Tool outputs\n- Audit logs\n- Context \u0026 memory snapshots\n\n---\n\n## 🧩 Example: AI Food Ordering with Approval\n1. Plan order\n2. Calculate total\n3. Generate checkout summary\n4. **Pause for approval**\n5. Browser agent completes checkout\n6. Persist confirmation \u0026 logs\n\n\u003cimg src=\"assets/human-approval.png\" title=\"SuperAgentX Architecture\"/\u003e\n\n## Browser AI Agent\n\n#### Install Playwright for Browser AI Automation\n```bash\npip install playwright\n```\n\n```bash\nplaywright install\n```\n## Example 1\n\n```python\nimport asyncio\nimport json\nfrom superagentx.agent import Agent\nfrom superagentx.agentxpipe import AgentXPipe\nfrom superagentx.browser_engine import BrowserEngine\nfrom superagentx.llm import LLMClient\nfrom superagentx.prompt import PromptTemplate\n\n\nasync def main():\n    print(\"SuperAgentX – Food Checkout \u0026 Payment Automation\")\n\n    # ------------------------------------------------------------------\n    # LLM SETUP\n    # ------------------------------------------------------------------\n    llm = LLMClient(\n        llm_config={\n            \"model\": \"gemini/gemini-3-pro-preview\",\n            \"temperature\": 0.1\n        }\n    )\n\n    # ------------------------------------------------------------------\n    # AGENT 1: FOOD \u0026 SNACKS CHECKOUT AGENT\n    # ------------------------------------------------------------------\n    checkout_system_prompt = \"\"\"\n    You are a food \u0026 snacks checkout agent. Simulate Food \u0026 Snacks Checkout with items.\n\n    Task:\n    - Select food and snack items\n    - Decide quantities\n    - Calculate total amount\n    - Prepare checkout summary for payment\n\n    Rules:\n    - DO NOT generate any payment or card details\n    - DO NOT mention CVV, card numbers, or expiry\n    - Output ONLY valid JSON\n\n    JSON Schema:\n    {\n      \"items\": [\n        {\n          \"name\": string,\n          \"category\": \"food | snack\",\n          \"quantity\": number,\n          \"price_per_unit\": number\n        }\n      ],\n      \"currency\": \"USD\",\n      \"total_amount\": number,\n      \"checkout_note\": string\n    }\n    \"\"\"\n\n    checkout_prompt = PromptTemplate(system_message=checkout_system_prompt)\n\n    checkout_agent = Agent(\n        name=\"Food Checkout Agent\",\n        role=\"Food \u0026 Snacks Checkout Planner\",\n        goal=\"Prepare checkout summary\",\n        llm=llm,\n        prompt_template=checkout_prompt,\n        max_retry=1\n    )\n\n    # ------------------------------------------------------------------\n    # AGENT 2: BROWSER REVIEW \u0026 PAY AGENT\n    # ------------------------------------------------------------------\n    browser_system_prompt = \"\"\"\n    You are a browser automation agent responsible for review and payment.\n\n    Input:\n    - You will receive a checkout summary JSON from the previous agent.\n\n    Target Payment Form URL:\n    https://superagentxai.github.io/payment-demo.github.io/\n\n    Task:\n    1. Review checkout summary (items \u0026 total) and MUST set the price from result.total_amount\n    2. Show checkout summary in the popup with price\n    3. Generate DUMMY credit card details for testing:\n       - 16-digit test card number\n       - Future expiry (MM/YY)\n       - 3-digit CVV\n       - Realistic cardholder name \u0026 address\n    4. Fill the payment form using generated card details\n    5. Submit the form\n\n\n    Rules:\n    - Change the Price value in the submit button with the actual amount from result.total_amount in USD:.\n    - Card details must be generated ONLY by you\n    - Use dummy/test card numbers only (e.g., 4111 1111 1111 1111)\n    - Do NOT persist card data\n    - Do NOT assume submission success\n    - Extract confirmation text ONLY if visible\n\n    Output JSON:\n    {\n      \"submission_status\": \"success | failed\",\n      \"reviewed_total_amount\": number,\n      \"confirmation_text\": string | null\n    }\n    \"\"\"\n\n    browser_prompt = PromptTemplate(system_message=browser_system_prompt)\n\n    browser_engine = BrowserEngine(\n        llm=llm,\n        prompt_template=browser_prompt,\n        headless=False  # set True in CI\n    )\n\n    browser_agent = Agent(\n        name=\"Review \u0026 Pay Agent\",\n        role=\"Browser Payment Executor\",\n        goal=\"Review checkout and pay using credit card\",\n        llm=llm,\n        human_approval=True,   # governance point\n        prompt_template=browser_prompt,\n        engines=[browser_engine],\n        max_retry=2\n    )\n\n    # ------------------------------------------------------------------\n    # PIPELINE: AGENT 1 → AGENT 2\n    # ------------------------------------------------------------------\n    pipe = AgentXPipe(\n        agents=[checkout_agent, browser_agent], # Sequence Agent Workflow\n        workflow_store=True\n    )\n\n    result = await pipe.flow(\n        query_instruction=\"Checkout food and snacks, then review and pay using credit card.\"\n    )\n\n    formatted_result = [\n        {\n            \"agent_name\": r.name,\n            \"agent_id\": r.agent_id,\n            \"goal_satisfied\": r.is_goal_satisfied,\n            \"result\": r.result\n        }\n        for r in result\n    ]\n\n    print(\" Final Result (Formatted JSON)\")\n    print(json.dumps(formatted_result, indent=2))\n    return\n\n\nif __name__ == \"__main__\":\n    asyncio.run(main())\n\n```\n\n## Example 2\n\n```python\nimport asyncio\n\nfrom superagentx.agent import Agent\nfrom superagentx.browser_engine import BrowserEngine\nfrom superagentx.llm import LLMClient\nfrom superagentx.prompt import PromptTemplate\n\n\nasync def main():\n    llm_client: LLMClient = LLMClient(llm_config={'model': 'gpt-4.1', 'llm_type': 'openai'})\n\n    prompt_template = PromptTemplate()\n\n    browser_engine = BrowserEngine(\n        llm=llm_client,\n        prompt_template=prompt_template,\n\n    )\n    query_instruction = (\"Which teams have won more than 3 FIFA World Cups, and which team is most likely to win the \"\n                         \"next one?\")\n\n    fifo_analyser_agent = Agent(\n        goal=\"Complete user's task.\",\n        role=\"You are a Football / Soccer Expert Reviewer\",\n        llm=llm_client,\n        prompt_template=prompt_template,\n        max_retry=1,\n        engines=[browser_engine]\n    )\n\n    result = await fifo_analyser_agent.execute(\n        query_instruction=query_instruction\n    )\n\n    print(result)\n\n\nasyncio.run(main())\n\n```\n## Run\n\u003cimg src=\"https://github.com/superagentxai/superagentx/blob/master/assets/superagentx_browser.gif\" title=\"Browser Engine\"/\u003e\n\n\n## Key Features\n\n🚀 **Open-Source Framework**: A lightweight, open-source AI framework built for multi-agent applications with Artificial General Intelligence (AGI) capabilities.\n\n🎯 **Goal-Oriented Multi-Agents**: This technology enables the creation of agents with retry mechanisms to achieve set goals. Communication between agents is Parallel, Sequential, or hybrid.\n\n🏖️ **Easy Deployment**: Offers WebSocket, RESTful API, and IO console interfaces for rapid setup of agent-based AI solutions.\n\n♨️ **Streamlined Architecture**: Enterprise-ready scalable and pluggable architecture. No major dependencies; built independently!\n\n📚 **Contextual Memory**: Uses SQL + Vector databases to store and retrieve user-specific context effectively.\n\n🧠 **Flexible LLM Configuration**: Supports simple configuration options of various Gen AI models.\n\n🤝 **Extendable Handlers**: Allows integration with diverse APIs, databases, data warehouses, data lakes, IoT streams, and more, making them accessible for function-calling features.\n\n💎 **Agentic RPA (Robotic Process Automation)** – SuperAgentX enables computer-use automation for both browser-based and desktop applications, making it an ideal solution for enterprises looking to streamline operations, reduce manual effort, and boost productivity.\n\n\n### Getting Started\n\n```shell\npip install superagentx\n```\n##### Usage - Example SuperAgentX Code\nThis SuperAgentX example utilizes two handlers, Amazon and Walmart, to search for product items based on user input from the IO Console.\n\n1. It uses Parallel execution of handler in the agent \n2. Memory Context Enabled\n3. LLM configured to OpenAI\n4. Pre-requisites\n\n## Environment Setup\n```shell\n$ python3.12 -m pip install poetry\n$ cd \u003cpath-to\u003e/superagentx\n$ python3.12 -m venv venv\n$ source venv/bin/activate\n(venv) $ poetry install\n```\n\n## [Documentation](https://docs.superagentx.ai/introduction)\n\n## License\n\nSuperAgentX is released under the [MIT](https://github.com/superagentxai/superagentX/blob/master/LICENSE) License.\n\n## 🤝 Contributing\nFork → Branch → Commit → Pull Request  \nKeep contributions modular and documented.\n\n## 📬 Connect\n- 🌐 https://www.superagentx.ai\n- 💻 https://github.com/superagentxai/superagentx\n\n⭐ Star the repo and join the community!\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsuperagentxai%2Fsuperagentx","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsuperagentxai%2Fsuperagentx","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsuperagentxai%2Fsuperagentx/lists"}