{"slug": "veryfront-code-the-first-full-stack-ai-framework", "title": "Veryfront Code, the first full-stack AI framework", "summary": "Veryfront released Veryfront Code, an open-source full-stack framework for building AI apps and agents with TypeScript and React, supporting Node.js, Deno, and Bun runtimes. The framework includes features such as agents, tools, workflows, and MCP server support, and can be self-hosted or used with Veryfront Cloud. Developers can start with the default ai-agent template targeting Node.js 22.3 or later, using direct OpenAI inference.", "body_md": "**Build full-stack AI apps and agents with TypeScript and React.**\n\nVeryfront Code is an open-source, full-stack framework that keeps React routes, agents, typed tools, evals, and multi-step workflows in one file-based TypeScript project. Developers and coding agents work from the same source and conventions.\n\nRun the same project on Node.js, Deno, or Bun. Connect directly to model providers or OpenAI-compatible local services.\n\n[Self-host](https://veryfront.com/docs/code/guides/self-hosting) without a\nVeryfront account. Use the [Veryfront Cloud AI Gateway](https://veryfront.com/docs/code/guides/providers#veryfront-cloud-ai-gateway)\nfor managed model access, and use [Veryfront Cloud](https://veryfront.com/docs/code/getting-started/cloud-quickstart)\nfor previews and production hosting.\n\nBuild agents, workflows, and full-stack React applications from one TypeScript project.\n\n-\n- Build AI agents that reason and act. Give them instructions, models, tools, skills, memory, and durable hosted execution.**Agents** -\n- Add project-level agent capabilities with**Skills**`SKILL.md`\n\nfiles. Skills package instructions, allowed tools, and scripts. -\n- Define Zod-validated functions that agents can call. Tools are discovered from files, so you do not need manual registration.**Tools** -\n- Reuse named prompt templates across agents, tools, MCP servers, and application code.**Prompts** -\n- Turn source documents into project knowledge files that agents can use as context.**Knowledge** -\n- Give agents conversation history, streamed responses, and React chat UI components with AG-UI support.**Memory & Streaming** -\n- Compose orchestrators and sub-agents that delegate to each other as tools for coordinated work.**Multi-Agent Systems** -\n- Define file-based background jobs that Veryfront Code discovers and runs as task executions.**Tasks** -\n- Orchestrate multi-step AI pipelines with branching, parallel steps, approval gates, and durable Redis checkpoints.**Workflows** -\n- Execute durable task, workflow, and agent work through project-scoped run records.**Runs** -\n- Expose tools, prompts, and resources through MCP with SSE transport, sessions, and elicitation.**MCP Server** -\n- Run isolated code in ephemeral compute environments with shell tools and agent-service integration.**Sandbox** -\n- Add third-party services with connectors for OAuth, remote tools, and service metadata.**Integrations** -\n- Build app routes with files, React Server Components, layouts, and server-side rendering.**Pages & Routing** -\n- Load server data, define API handlers, and add**Data Fetching & API Routes**[middleware](https://veryfront.com/docs/code/guides/middleware)with built-in[OAuth](https://veryfront.com/docs/code/guides/oauth). -\n- Extend Veryfront Code with contract-based packages for LLM providers, bundling, CSS, tracing, caching, and more.**Extensions**\n\nThe default `ai-agent`\n\nstarter targets Node.js 22.3 or later. This path uses\ndirect OpenAI inference. Create the project, configure inference, and start the\ndevelopment server:\n\n```\nnpm create veryfront@latest my-agent -- --template ai-agent\ncd my-agent\nexport OPENAI_API_KEY=\"<API_KEY>\"\nnpm run dev\n```\n\nOpen the URL printed by the CLI and ask `What is 128 divided by 8?`\n\n. The agent\ncalls the starter's calculator tool and returns `16`\n\n.\n\nTo use the Veryfront Cloud AI Gateway, another provider, an OpenAI-compatible\nlocal service, or the optional ONNX extension, [select another inference path](https://veryfront.com/docs/code/guides/providers)\nbefore the first request.\n\nNode.js is the default runtime. Select Deno or Bun explicitly:\n\n```\nnpx veryfront@latest init <PROJECT_NAME> --template ai-agent --runtime deno\nnpx veryfront@latest init <PROJECT_NAME> --template ai-agent --runtime bun\n```\n\nFollow the [Quickstart guide](https://veryfront.com/docs/code/getting-started/quickstart)\nfor the complete first project, including its smoke eval. See\n[Create a project](https://veryfront.com/docs/code/getting-started/create-project)\nfor every setup option.\n\n## Use another package manager\n\nThese commands open the project wizard:\n\n```\npnpm create veryfront\nyarn create veryfront\nbun create veryfront\ndeno init --npm veryfront\n```\n\n## Choose another starter\n\nChoose a starter with `--template`\n\n:\n\n```\nnpx veryfront@latest init <PROJECT_NAME> --template <TEMPLATE>\n```\n\n| Starter | Starting point |\n|---|---|\n`ai-agent` |\nAgent, chat UI, tools, and streaming |\n`minimal` |\nBlank full-stack application |\n`docs-agent` |\nDocument Q&A with source citations |\n`agentic-workflow` |\nSteps, approvals, and parallelism |\n`multi-agent-system` |\nAgents that delegate to each other |\n`coding-agent` |\nCode assistant with file tools |\n`saas-starter` |\nAuthentication, chat, and user memory |\n\n## Install the CLI globally\n\nInstall through npm for local development commands and the TUI:\n\n```\nnpm install -g veryfront@latest\n```\n\nThe standalone binary includes the runtime. Downloads range from 0.9 to 1.2 GB by platform:\n\n```\ncurl -fsSL https://veryfront.com/install.sh | sh\n# or\nbrew install veryfront/tap/veryfront\n```\n\nVeryfront projects use conventional directories:\n\n```\napp/          React pages, layouts, and API routes\nagents/       Agent definitions\ntools/        Typed tools\nprompts/      Prompt templates\nskills/       SKILL.md instruction packs\nevals/        Agent and workflow evals\nworkflows/    Multi-step workflow DAGs\ntasks/        Background work targets\nschedules/    Scheduled run creation\nwebhooks/     Webhook triggers\nresources/    MCP resources\n```\n\nVeryfront discovers these directories at startup. Schedules and webhooks trigger agent runs, workflow runs, or task runs.\n\nEvery starter includes `AGENTS.md`\n\nwith project guidance. While `veryfront dev`\n\nruns, MCP-aware coding agents can use Veryfront MCP tools for live errors, route\ninspection, scaffolding, tests, linting, and HMR. See\n[Coding agents](https://veryfront.com/docs/code/guides/coding-agents) and the\n[project structure guide](https://veryfront.com/docs/code/guides/project-structure).\n\nRead [CONTRIBUTING.md](/veryfront/veryfront-code/blob/main/CONTRIBUTING.md) before contributing. For code\nchanges, open an issue before opening a pull request.\n\nUse GitHub issues for reproducible bugs and feature requests. Join the\n[community Discord](https://discord.gg/xWuRjafrtV) for setup questions and\nproject discussion.\n\nReport security vulnerabilities privately to\n[security@veryfront.com](mailto:security@veryfront.com). Veryfront responds\nwithin 48 hours.\n\nApache-2.0", "url": "https://wpnews.pro/news/veryfront-code-the-first-full-stack-ai-framework", "canonical_source": "https://github.com/veryfront/veryfront-code", "published_at": "2026-08-26 08:11:26+00:00", "updated_at": "2026-08-26 08:45:31.828837+00:00", "lang": "en", "topics": ["ai-tools", "developer-tools", "ai-agents", "ai-infrastructure"], "entities": ["Veryfront", "Veryfront Code", "OpenAI", "Node.js", "Deno", "Bun", "React", "TypeScript"], "alternates": {"html": "https://wpnews.pro/news/veryfront-code-the-first-full-stack-ai-framework", "markdown": "https://wpnews.pro/news/veryfront-code-the-first-full-stack-ai-framework.md", "text": "https://wpnews.pro/news/veryfront-code-the-first-full-stack-ai-framework.txt", "jsonld": "https://wpnews.pro/news/veryfront-code-the-first-full-stack-ai-framework.jsonld"}}