{"id":49023140,"url":"https://github.com/aaronrussell/agentflow","last_synced_at":"2026-04-19T05:01:36.773Z","repository":{"id":257805559,"uuid":"838488888","full_name":"aaronrussell/agentflow","owner":"aaronrussell","description":"Powerfully simple AI agent framework.","archived":false,"fork":false,"pushed_at":"2025-01-24T17:47:09.000Z","size":1045,"stargazers_count":42,"open_issues_count":1,"forks_count":4,"subscribers_count":2,"default_branch":"main","last_synced_at":"2026-03-26T18:40:34.881Z","etag":null,"topics":["agents","ai","cli","markdown"],"latest_commit_sha":null,"homepage":"https://agentflow.2point0.ai","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aaronrussell.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}},"created_at":"2024-08-05T18:38:59.000Z","updated_at":"2026-01-28T18:23:22.000Z","dependencies_parsed_at":null,"dependency_job_id":"22653626-d158-4af2-a02b-310f6ad7d274","html_url":"https://github.com/aaronrussell/agentflow","commit_stats":null,"previous_names":["lebrunel/agentflow","aaronrussell/agentflow"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/aaronrussell/agentflow","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aaronrussell%2Fagentflow","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aaronrussell%2Fagentflow/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aaronrussell%2Fagentflow/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aaronrussell%2Fagentflow/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aaronrussell","download_url":"https://codeload.github.com/aaronrussell/agentflow/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aaronrussell%2Fagentflow/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31995167,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-18T20:23:30.271Z","status":"online","status_checked_at":"2026-04-19T02:00:07.110Z","response_time":55,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"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":["agents","ai","cli","markdown"],"created_at":"2026-04-19T05:01:14.621Z","updated_at":"2026-04-19T05:01:36.758Z","avatar_url":"https://github.com/aaronrussell.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Agentflow\n\n![Agentflow](https://raw.githubusercontent.com/lebrunel/agentflow/main/media/agentflow-banner.webp)\n\n![GitHub License](https://img.shields.io/github/license/lebrunel/agentflow?style=flat-square)\n![Build Status](https://img.shields.io/github/actions/workflow/status/lebrunel/agentflow/bun.yml?style=flat-square)\n\nAgentflow is a low-code framework for creating and executing AI-powered workflows using Markdown and natural language.\n\n- 💬 **Natural language as code** - Write powerful workflows using natural language and Markdown, making automation accessible to all.\n- 🧠 **AI-powered execution** - Take advantage of cutting-edge AI models to bring super-intelligence to your workflows.\n- 🔌 **Flexible and adaptable** - Connect with any AI provider or run models locally, for complete control and flexibility.\n- 🔀 **Full logical control** - Create complex workflows with loops and conditional branching, just like traditional programming but in plain English.\n- 🛠️ **Extend with ease** - Add custom actions and tools using JavaScript to tailor Agentflow to your specific needs.\n\n\u003e [!WARNING]\n\u003e Agentflow is in early development. While it's ready for experimentation and early adoption, the API may change as we continue to improve the framework. Always refer to the [documentation](https://agentflow.2point0.ai) for the latest updates.\n\n## Table of contents\n\n- [Quick start](#quick-start)\n- [Writing workflows](#writing-workflows)\n- [JavaScript API](#javascript-api)\n- [Documentation](#documentation)\n- [License](#license)\n\n## Quick start\n\nAgentflow requires a JavaScript runtime (for example, Node.js) to be installed on your system. Once you have that set up, there are two simple ways to get started with Agentflow: using the starter kit to bootstrap a project, or using the CLI to initialize a project manually.\n\n### 1. Using the Agentflow starter kit\n\nThe quickest way to get started is using the Agentflow starter kit. Choose your preferred package manager:\n\n```sh\n# Using NPM\nnpm create agentflow@latest my-agents\n\n# Using Yarn\nyarn create agentflow my-agents\n```\n\nThis will scaffold a new project with all the necessary dependencies and configuration. The CLI will be installed locally as a project dependency, which you can use via your package manager:\n\n```sh\n# Using NPM\ncd my-agents \u0026\u0026 npm install\nnpx aflow help\n\n# Using Yarn\ncd my-agents \u0026\u0026 yarn\nyarn aflow help\n```\n\n### 2. Using the global CLI\n\nAlternatively, you can install the Agentflow CLI globally and initialize a new project via the CLI:\n\n```sh\n# Install CLI globally\nnpm install -g @agentflow/cli\n\n# Create new project\naflow init my-agents\ncd my-agents\n```\n\nFor more detailed instructions and configuration options, check out the [getting started guide](https://agentflow.2point0.ai/guide/getting-started) and [CLI documentation](https://agentflow.2point0.ai/guide/cli).\n\n## Writing workflows\n\nWorkflows in Agentflow are written using Markdown files that combine natural language instructions with actions using MDX-like syntax. Actions can do anything from generating text with AI models to implementing control flow and integrating external services.\n\nHere's a simple example:\n\n```mdx\n---\ndata:\n  languages:\n    - Spanish\n    - French\n    - German\ninput:\n  topic:\n    type: text\n    message: \"Enter a topic to write about\"\n---\n\nWrite a short blog post about {topic}.\n\n\u003cGenText as=\"article\" model=\"openai:gpt-4o\" /\u003e\n\n\u003cLoop\n  as=\"translations\"\n  until={$.index === languages.length}\n  provide={{\n    original: article,\n    language: languages[$.index],\n  }}\u003e\n\n  Translate this article to {language}:\n  {original}\n\n  \u003cGenText as=\"translated\" model=\"openai:gpt-4o\" /\u003e\n\u003c/Loop\u003e\n```\n\nThis workflow prompts for a topic, generates a blog post, then automatically translates it into multiple languages. Actions like `\u003cGenText /\u003e` invoke AI models to generate content, while control flow actions like `\u003cLoop /\u003e` enable complex workflows. Expressions in braces (e.g. `{topic}`) reference previous inputs and results.\n\nCheck out the docs to learn more about [workflow structure](https://agentflow.2point0.ai/guide/workflow-structure) and [working with input data](https://agentflow.2point0.ai/guide/input-data).\n\n## JavaScript API\n\nAgentflow provides a JavaScript API for working with workflows programmatically. Using `@agentflow/core`, you can compile Markdown-formatted workflows and control their execution from your code:\n\n```typescript\nimport { Environment, Workflow } from '@agentflow/core'\n\n// Initialize the environment\nconst env = new Environment({\n  providers: {\n    // configure AI providers\n  }\n})\n\n// Compile a workflow from markdown\nconst workflow = Workflow.compileSync(`\n  Write a haiku about {topic}.\n\n  \u003cGenText as=\"haiku\" model=\"openai:gpt-4o-mini\" /\u003e\n`, env)\n\n// Create and run the execution\nconst ctrl = workflow.createExecution({\n  topic: { type: 'primitive', value: 'spring rain' }\n})\n\n// Handle execution events\nctrl.on('step', (step, event) =\u003e {\n  event.action?.then(({ result }) =\u003e console.log(result))\n})\n\nctrl.on('complete', (output) =\u003e {\n  console.log('Done!\\n', output)\n})\n\n// Run the workflow\nawait ctrl.runAll()\n```\n\nThe API provides fine-grained control over workflow execution, including the ability to pause, resume and monitor progress. Check out the [JavaScript API documentation](https://agentflow.2point0.ai/guide/javascript-api) for more details.\n\n## Documentation\n\nFor more detailed guides, examples and API documentation, visit [agentflow.2point0.ai](https://agentflow.2point0.ai).\n\n## License\n\nThis package is open source and released under the [Apache-2 Licence](https://github.com/lebrunel/agentflow/blob/master/LICENSE).\n\n© Copyright 2024 [Push Code Ltd](https://www.pushcode.com/).\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faaronrussell%2Fagentflow","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faaronrussell%2Fagentflow","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faaronrussell%2Fagentflow/lists"}