[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72205":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":27,"readmeContent":28,"aiSummary":29,"trendingCount":16,"starSnapshotCount":16,"syncStatus":17,"lastSyncTime":30,"discoverSource":31},72205,"workflow-use","browser-use\u002Fworkflow-use","browser-use","⚙️ Create and run workflows (RPA 2.0)","https:\u002F\u002Fbrowser-use.com",null,"Python",4045,330,30,35,0,2,13,41,6,74.16,"GNU Affero General Public License v3.0",false,"main",true,[],"2026-06-12 04:01:04","\u003Cpicture>\n  \u003Cimg alt=\"Workflow Use logo - a product by Browser Use.\" src=\".\u002Fstatic\u002Fworkflow-use.png\"  width=\"full\">\n\u003C\u002Fpicture>\n\n\u003Cbr \u002F>\n\n\u003Ch1 align=\"center\">Deterministic, Self Healing Workflows (RPA 2.0)\u003C\u002Fh1>\n\n[![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fbrowser-use\u002Fworkflow-use?style=social)](https:\u002F\u002Fgithub.com\u002Fbrowser-use\u002Fworkflow-use\u002Fstargazers)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1303749220842340412?color=7289DA&label=Discord&logo=discord&logoColor=white)](https:\u002F\u002Flink.browser-use.com\u002Fdiscord)\n[![Cloud](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FCloud-☁️-blue)](https:\u002F\u002Fcloud.browser-use.com)\n[![Twitter Follow](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002FGregor?style=social)](https:\u002F\u002Fx.com\u002Fgregpr07)\n[![Twitter Follow](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002FMagnus?style=social)](https:\u002F\u002Fx.com\u002Fmamagnus00)\n\n⚙️ **Workflow Use** is the easiest way to create and execute deterministic workflows with variables which fallback to [Browser Use](https:\u002F\u002Fgithub.com\u002Fbrowser-use\u002Fbrowser-use) if a step fails. You just _show_ the recorder the workflow, we automatically generate the workflow.\n\n❗ This project is in very early development so we don't recommend using this in production. Lots of things will change and we don't have a release schedule yet. Originally, the project was born out of customer demand to make Browser Use more reliable and deterministic.\n\n## 🚀 NEW: Generation Mode\n\nAutomatically generate workflows from natural language! Describe your task, we run browser-use once, then create a reusable semantic workflow stored in a database.\n\n### Quick Commands\n\n```bash\n# Generate workflow from task description\npython cli.py generate-workflow \"Find GitHub stars for browser-use repo\"\n\n# List all workflows\npython cli.py list-workflows\n\n# Filter by generation mode\npython cli.py list-workflows --generation-mode browser_use\n\n# Run stored workflow\npython cli.py run-stored-workflow \u003Cworkflow-id> --prompt \"Find stars for playwright repo\"\n\n# View workflow details\npython cli.py workflow-info \u003Cworkflow-id>\n\n# Delete workflow\npython cli.py delete-workflow \u003Cworkflow-id>\n```\n\n### How It Works\n\n1. **Describe**: Give a task in natural language\n2. **Execute**: Browser-use completes the task once\n3. **Generate**: Execution history → semantic workflow with parameters\n4. **Store**: Save to database with metadata\n5. **Reuse**: Run the workflow with different inputs, no AI needed\n\n### Advanced Options\n\n```bash\n# Custom models for generation\npython cli.py generate-workflow \"Your task\" \\\n  --agent-model \"gpt-4.1-mini\" \\\n  --extraction-model \"gpt-4.1-mini\" \\\n  --workflow-model \"gpt-4o\"\n\n# Use Browser-Use Cloud browser\npython cli.py generate-workflow \"Your task\" --use-cloud\n\n# Save to custom location\npython cli.py generate-workflow \"Your task\" --output-file .\u002Fmy-workflow.json\n\n# Skip database storage\npython cli.py generate-workflow \"Your task\" --no-save-to-storage\n```\n\n### Storage\n\nWorkflows stored at `workflows\u002Fstorage\u002F`:\n- `metadata.json` - Searchable index of all workflows\n- `workflows\u002F\u003Cid>.workflow.json` - Individual workflow files\n\n### Programmatic Usage\n\n```python\nfrom workflow_use.healing.service import HealingService\nfrom workflow_use.storage.service import WorkflowStorageService\nfrom browser_use.llm import ChatOpenAI\n\nhealing_service = HealingService(llm=ChatOpenAI(model='gpt-4.1'))\nstorage_service = WorkflowStorageService()\n\n# Generate workflow\nworkflow = await healing_service.generate_workflow_from_prompt(\n    prompt=\"Fill contact form on example.com\",\n    agent_llm=ChatOpenAI(model='gpt-4.1-mini'),\n    extraction_llm=ChatOpenAI(model='gpt-4.1-mini'),\n    use_cloud=True  # Optional: use Browser-Use Cloud\n)\n\n# Save to storage\nmetadata = storage_service.save_workflow(\n    workflow=workflow,\n    generation_mode='browser_use',\n    original_task=\"Fill contact form on example.com\"\n)\n\n# Retrieve and execute\nloaded_workflow = storage_service.get_workflow(metadata.id)\n```\n\n# Quick start\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fbrowser-use\u002Fworkflow-use\n```\n\n## Build the extension\n\n```bash\ncd extension && npm install && npm run build\n```\n\n## Setup workflow environment\n\n```bash\ncd .. && cd workflows\nuv sync\nsource .venv\u002Fbin\u002Factivate # for mac \u002F linux\nplaywright install chromium\ncp .env.example .env # add your OPENAI_API_KEY to the .env file\n```\n\n\n## Run workflow as tool\n\n```bash\npython cli.py run-as-tool examples\u002Fexample.workflow.json --prompt \"fill the form with example data\"\n```\n\n## Run workflow with predefined variables\n\n```bash\npython cli.py run-workflow examples\u002Fexample.workflow.json\n```\n\n## Record your own workflow\n\n```bash\npython cli.py create-workflow\n```\n\n## See all commands\n\n```bash\npython cli.py --help\n```\n\n# Usage from python\n\nRunning the workflow files is as simple as:\n\n```python\nfrom workflow_use import Workflow\n\nworkflow = Workflow.load_from_file(\"example.workflow.json\")\nresult = asyncio.run(workflow.run_as_tool(\"I want to search for 'workflow use'\"))\n```\n\n## Cloud Browser Support\n\nRun workflows in [Browser-Use Cloud](https:\u002F\u002Fcloud.browser-use.com) (new signups get $10 free credits via OAuth or $1 via email) with semantic abstraction (no AI):\n(NOTE: Set BROWSER_USE_API_KEY environment variable)\n```python\nfrom workflow_use import Workflow\n\nworkflow = Workflow.load_from_file(\"workflow.json\", llm, use_cloud=True)\nresult = await workflow.run_with_no_ai()  # No LLM calls, uses semantic mapping\n```\n\nExamples:\n- `examples\u002Fcloud_browser_demo.py` - Load recorded workflow and run on cloud\n\n## Launch the GUI\n\nThe Workflow UI provides a visual interface for managing, viewing, and executing workflows.\n\n### Option 1: Using the CLI command (Recommended)\n\nThe easiest way to start the GUI is with the built-in CLI command:\n\n```bash\ncd workflows\npython cli.py launch-gui\n```\n\nThis command will:\n- Start the backend server (FastAPI)\n- Start the frontend development server\n- Automatically open http:\u002F\u002Flocalhost:5173 in your browser\n- Capture logs to the `.\u002Ftmp\u002Flogs` directory\n\nPress Ctrl+C to stop both servers when you're done.\n\n### Option 2: Start servers separately\n\nAlternatively, you can start the servers individually:\n\n#### Start the backend server\n\n```bash\ncd workflows\nuvicorn backend.api:app --reload\n```\n\n#### Start the frontend development server\n\n```bash\ncd ui\nnpm install\nnpm run dev\n```\n\nOnce both servers are running, you can access the Workflow GUI at http:\u002F\u002Flocalhost:5173 in your browser. The UI allows you to:\n\n- Visualize workflows as interactive graphs\n- Execute workflows with custom input parameters\n- Monitor workflow execution logs in real-time\n- Edit workflow metadata and details\n\n# Demos\n\n## Workflow Use filling out form instantly\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fcf284e08-8c8c-484a-820a-02c507de11d4\n\n## Gregor's explanation\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F379e57c7-f03e-4eb9-8184-521377d5c0f9\n\n# Features\n\n- 🔁 **Record Once, Reuse Forever**: Record browser interactions once and replay them indefinitely.\n- ⏳ **Show, don't prompt**: No need to spend hours prompting Browser Use to do the same thing over and over again.\n- ⚙️ **Structured & Executable Workflows**: Converts recordings into deterministic, fast, and reliable workflows which automatically extract variables from forms.\n- 🪄 **Human-like Interaction Understanding**: Intelligently filters noise from recordings to create meaningful workflows.\n- 🔒 **Enterprise-Ready Foundation**: Built for future scalability with features like self-healing and workflow diffs.\n\n# Vision and roadmap\n\nShow computer what it needs to do once, and it will do it over and over again without any human intervention.\n\n## Workflows\n\n- [ ] Nice way to use the `.json` files inside python code\n- [ ] Improve LLM fallback when step fails (currently really bad)\n- [ ] Self healing, if it fails automatically agent kicks in and updates the workflow file\n- [ ] Better support for LLM steps\n- [ ] Take output from previous steps and use it as input for next steps\n- [ ] Expose workflows as MCP tools\n- [ ] Use Browser Use to automatically create workflows from websites\n\n## Developer experience\n\n- [ ] Improve CLI\n- [ ] Improve extension\n- [ ] Step editor\n\n## Agent\n\n- [ ] Allow Browser Use to use the workflows as MCP tools\n- [ ] Use workflows as website caching layer\n","Workflow Use 是一个用于创建和运行确定性工作流（RPA 2.0）的工具。其核心功能包括通过自然语言描述任务自动生成工作流，并在步骤失败时自动回退到Browser Use以提高可靠性。项目使用Python开发，支持从命令行界面执行生成、列出、运行、查看和删除工作流等操作，同时提供高级选项如自定义模型和云浏览器使用。尽管目前仍处于早期开发阶段，不适合生产环境使用，但它非常适合需要自动化网页操作且希望流程更加稳定可靠的场景，例如数据抓取、表单填写等任务。","2026-06-11 03:40:51","high_star"]