Ever Works Platform
🌟 What is it
Ever® Works™ - The Workshop for AI - an open-source, extensible, agentic runtime that autonomously researches, ships, and maintains entire businesses, 24/7.
Give it an idea, and AI agents handle the rest:
- 🔬 Autonomous research — agents investigate the topic before building anything
- ✍️ AI content generation — long-form, structured, ready-to-publish content
- 💻 Code generation — full sites and repositories, not just snippets
- 🚀 Multi-target deployment — ship the result to your hosting of choice
- 🗂️ Git-native lifecycle — code and content both live in Git, so you own everything and nothing is locked in
- ⏰ Continuous improvement — keeps working on a schedule, not just when you prompt it
Think of it as a workshop for AI: it runs on your machine, your cloud, or ours, and keeps the whole content-and-code lifecycle version-controlled from the first idea to every future update.
Ever® Works™ is a part of Ever® Platform™ - The everything platform for businesses™.
✨ Features
Main capabilities:
- Idea-to-site automation — turn a single prompt into a researched, written, coded, and deployed result
- Autonomous AI agents with a configurable, multi-step generation pipeline
- Content management — items, categories, tags, and collections
- Code + content in Git — every change is a commit, fully version-controlled and portable
- Multi-provider AI — bring your own models and keys (OpenAI, Anthropic, Google, Groq, Mistral, Ollama, and more) via gateways like OpenRouter
- Extensible plugin system — AI providers, search, content extraction, screenshots, Git providers, deployment, and pipelines (see Plugins & Extensibility)
- Scheduled updates — keep generated sites fresh and improving over time
- Data import — bring in existing content from external sources
- Community PR workflow — contribute generated changes back through Git pull requests
- Comparison generator — produce structured comparison content
- Headless REST API — automate everything programmatically
- AI chat / conversations — interact with your works through natural language
- CLI — drive generation and operations from the terminal
- MCP server — expose the platform to AI agents and IDEs via the Model Context Protocol
- Auth — email/password plus OAuth (GitHub, Google)
- Subscriptions & billing, notifications, email, activity logs, and integrations
- Monitoring — built-in Sentry + PostHog instrumentation
- Multi-app, multi-surface — Web UI, API, CLI, Admin, MCP, and Docs in a single monorepo
Read more in the official documentation.
🧩 Plugins & Extensibility
Ever Works ships with a first-class plugin system. Plugins are standalone ESM packages that declare their metadata, capabilities, and settings — so you can mix, match, and extend providers without forking the core.
| Category | Plugins |
|---|
| AI providers | OpenAI, Anthropic, Google, Groq, Mistral, Ollama |
| AI gateways | OpenRouter (default), Vercel AI Gateway |
| Search | Tavily (default), Brave, Exa, SerpAPI, Perplexity, Bright Data, Firecrawl, Jina, Valyu, Linkup |
| Content extractors | Local extractor (default), Notion, PDF, Scrapfly |
| Screenshots | ScreenshotOne, Urlbox, Scrapfly |
| Git providers | GitHub (default, with OAuth) |
| Deployment | Vercel (default) |
| Data sources | Apify |
| Pipelines | Standard 15-step (default), Agent pipeline, Claude Code, Claude Managed Agent, Codex, Gemini, OpenCode, Make, Sim AI, Zapier |
| Prompt providers | Langfuse |
| Utilities | Comparison generator |
See the plugin documentation for how to build and publish your own.
🌼 Screenshots
Web UI
🔗 Links
📊 Activity
💻 Demo, Testing and Production
Demo
Try the Ever Works Platform live — no signup required. Head to https://ever.works, type your prompt, pick the type of work you want it to do, and watch the AI agents research, write, and build it in real-time.
See an example of what it builds. A content-rich website (directory) generated by Ever Works is live at https://demo.ever.works — again, no credentials needed, just explore.
Notes:
- The demo directory at https://demo.ever.works is generated output, not the platform UI itself — to see the platform in action, use https://ever.works as described above.
- The demo directory's content may be refreshed on each deployment (usually daily) via CI/CD from the
develop branch.
Production (SaaS)
Ever® Works™ Platform SaaS is available at https://app.ever.works.
Note: it's currently in pre-release/testing mode, please use it cautiously!
Staging
- Staging builds (using CI/CD, from the
stage branch) are used to test releases before they are deployed to production.
🧱 Technology Stack and Requirements
For Production, we recommend:
See also the README.md and LICENSES.md files in relevant folders for lists of libraries and software included in the Platform, information about licenses, and other details
🧰 Apps & Packages
Ever Works is a Turborepo + pnpm monorepo.
Apps (apps/*):
api — NestJS REST API (port 3100)web — Next.js App Router web UI (port 3000)cli — public CLIinternal-cli — internal NestJS CLI toolingadmin — admin interfacemcp — Model Context Protocol server (port 3200)docs — documentation site (renders docs/)
Packages (packages/*):
agent — core AI agent logic (generators, pipeline, work operations, facades, …)plugin — plugin system contracts & utilitiesplugins/* — provider implementations (AI, search, deployment, …)contracts — shared TypeScript typestasks — Trigger.dev background jobsmonitoring — Sentry + PostHog integrationcli-shared — shared CLI utilities
📄 Documentation
Please refer to our official Platform Documentation.
🚀 Quick Start
With Docker Compose
Demo
Production
- Edit
.env.compose (if needed) to use your custom settings.
- Run
docker compose up -d, if you want to run the platform in a minimal production configuration using our prebuilt Docker images.
Note: we recommend using Kubernetes for production workloads instead of Docker Compose!
Build
- Run
docker compose -f docker-compose.build.yml up -d, if you want to build everything (code and Docker images) locally. (Note: this builds the whole platform locally and can take a while — the options above are much faster.)
Notes:
- You can run only the infra dependencies (PostgreSQL + Redis, without our API / Web containers) with
docker compose -f docker-compose.infra.yml up -d. This is handy when running the apps locally while keeping the backing services in containers.
- You can add
--env-file .env.something to the docker compose command to use a specific env file with your custom settings.
Manually
Required
- Install Node.js LTS (>= 22).
- Install pnpm (>= 9.9).
- Install dependencies with
pnpm install.
- If you will make code changes (and push to the Git repo), run
pnpm prepare:husky.
- Copy the example env files and adjust settings:
apps/api/.env.example → apps/api/.envapps/web/.env.example → apps/web/.env.local
- Build workspace packages first (Turborepo handles dependency order):
pnpm build.
- Run the apps:
pnpm dev:apps — run all apps in watch modepnpm dev:api — API only (port 3100)pnpm dev:web — Web only (port 3000)
- Open http://localhost:3000 in your browser.
- Enjoy!
Notes:
- The API self-applies pending database migrations on startup, so there is nothing to run manually on a fresh setup.
- For Redis and PostgreSQL during local development, you can use
docker compose -f docker-compose.infra.yml up -d.
Production
Kubernetes
- We recommend deploying to Kubernetes (k8s) for production workloads. See the
.deploy/ folder for our deployment configurations.
Anonymous usage statistics
Anonymous usage statistics are off by default for now: a self-hosted installation sends nothing unless you opt in. Switch them on with EVER_STATS_ENABLED=true (in .env.compose with Docker Compose, in apps/api/.env when you run the API from source, or in the container environment). Once on, the installation sends a small, signed, anonymous report once a day — and on days 1–3 of a month one more, for the month that just closed — with counts, monthly totals and feature switches only, never a name, an address, a URL or any of your content — so the maintainers can see which versions and features are in use. You can switch it off again with EVER_STATS_ENABLED=false or in Settings → Ever Platform, where you can also see every byte that was sent. See Anonymous usage statistics.
☁️ One-Click Deploy
💌 Contact Us
🔐 Security
Ever® Works™ follows good security practices, but 100% security cannot be guaranteed in any software!
Ever® Works™ is provided AS IS without any warranty. Use at your own risk!
See more details in the LICENSE.
In a production setup, all client-side to server-side (backend, APIs) communications should be encrypted using HTTPS/WSS/SSL.
If you discover any issue regarding security, please disclose the information responsibly by sending an email to mailto:[email protected] and not by creating a GitHub issue.
🛡️ License
We support the open-source community. If you're building awesome non-profit/open-source projects, we're happy to help — feel free to contact us at mailto:[email protected] to make a request.
This software is available under the GNU Affero General Public License v3.0 (AGPL-3.0).
Please see LICENSE and LICENSES.md for more information on licenses
™️ Trademarks
Ever® is a registered trademark of Ever Co. LTD.
Ever® Demand™, Ever® Gauzy™, Ever® Teams™, Ever® Rec™, Ever® Recu™, Ever® Cloc™, Ever® Works™ and Ever® OpenSaaS™ are all trademarks of Ever Co. LTD.
The trademarks may only be used with the written permission of Ever Co. LTD. and may not be used to promote or otherwise market competitive products or services.
All other brand and product names are trademarks, registered trademarks, or service marks of their respective holders.
🍺 Contribute
- Please give us a :star: on Github, it helps!
- You are more than welcome to submit feature requests in the separate repo.
- Pull requests are always welcome! Please base pull requests against the develop branch and follow the contributing guide.
Automated contributions
Some pull requests in this repository are opened by Ever Works Fleet agents.
They come from branches named task/<slug>-<id>, and their commits are authored as <agent> <[email protected]>.
They target the develop branch and are never merged automatically: they get the same review and CI as any other pull request.
See the Fleet documentation for more details.
💪 Thanks to our Contributors
This project was initially developed in a private repository before its public release.
We are grateful to the current and former team members who contributed to the project during that phase.
Private-development contributors:
- Ruslan Konviser (989 commits, 1,031,214++ 191,454--)
- Claude (790 commits, 540,594++ 102,992--)
- Paradoxe Ng (273 commits, 569,521++ 263,134--)
- Gabriel Nt (63 commits, 124,405++ 37,680--)
- Dependabot (30 commits, 24,912++ 4,519--)
- Joel K. (26 commits, 16,373++ 7,214--)
- Michał D. (13 commits, 18,944++ 6,930--)
- Akilimaili C. I. (11 commits, 43,132++ 700--)
- Roland MN (5 commits, 14,063++ 155--)
Public contributions are tracked in the repository history from the initial public release onward in CONTRIBUTORS.md.
You can also view a full list of our contributors tracked by GitHub.
⭐ Star History
❤️ Powered By
©️ Copyright
Copyright © 2025-present, Ever Co. LTD. All rights reserved
🔥 P.S
