Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Windmill

Rust workflow engine that turns Python, TypeScript and SQL scripts into APIs, flows, cron jobs and internal apps, with AI agent steps.

Reviewed by ReviewRamble editors · Updated

View on GitHub Open source

Share this page:XLinkedInFacebookRedditWhatsAppEmail

Our rating

4.3/5

A fast, code-first workflow engine that suits engineering teams well, provided someone reads the licence and hardens the Docker defaults before it goes live.

What is Windmill?

Windmill runs scripts written in Python, TypeScript, Go, Bash, SQL and more than 20 languages in all, and turns each one into a generated form, an HTTP endpoint and a scheduled job. Scripts chain into flows built in a visual editor, as YAML or as plain code, and internal apps can be built on top of them in React or Svelte.

AI arrives in three ways: an in-browser AI chat and AI sessions that write scripts, flows and apps in a forked workspace for review; AI agent steps inside flows; and an MCP server that lets coding agents run jobs in a workspace. You bring your own model provider.

The backend is Rust, with stateless API servers and workers pulling jobs from a Postgres queue. Windmill Labs sells a cloud service and an Enterprise Edition; the repository dates from May 2022 and v1.876.1 shipped on 17 November 2026.

Is Windmill worth it?

Windmill is the strongest self-hosted choice when the people building automations are engineers who would rather write a Python or TypeScript function than drag boxes on a canvas. Each script gets a form, an endpoint and a schedule for free, flows can be written as code with approval steps that release the worker while waiting, and Git sync, a CLI and a VS Code extension keep the work in a repository. The project has about 18,600 GitHub stars and ships almost daily; v1.876.1 arrived on 17 November 2026.

The licence is more involved than the AGPLv3 badge suggests, and GitHub reports it as NOASSERTION for that reason. Code built without the enterprise compile flag is AGPLv3; the API clients and the OpenAPI spec are Apache 2.0; enterprise code is proprietary. The prebuilt Community Edition images include some of that proprietary code, are free to use internally without limit beyond in-product quotas, and may not be resold, run as a managed service or embedded in a product without a commercial licence.

Those quotas are where the free edition pinches. According to the pricing page it allows 50 users, 3 workspaces, 4 groups, Git sync for 2 users, 10 GiB of workspace storage, SSO for only 10 users and 30 days of job details. Audit logs, SAML and SCIM, external secret stores, autoscaling, concurrency limits and Kafka or SQS triggers are Enterprise features. The calculator's default quote is $170 a month for one builder, two 2 GB workers and eight native subworkers, and it counts only users active in the last 30 days.

The catch: the Docker Compose quickstart is a lab setup. It logs you in as admin@windmill.dev with the password changeme, uses changeme for Postgres, publishes plain HTTP on port 80 and SMTP on port 25, and runs the default workers privileged. The security docs say job isolation is off in the bare binary; the Compose file turns on PID namespace isolation for the default workers but not the native worker. GitHub lists five security advisories published in 2026, two rated high.

Our take: for teams already writing glue scripts, Windmill is a better buy than n8n, since it charges per seat and worker rather than per execution and the AI builder produces ordinary reviewed code. Teams whose builders are not programmers will find n8n or Activepieces easier going.

Where it earns its place

  • A data engineer at a 60-person logistics company

    Replace a box of cron scripts with something that has logs, retries and a UI

    Paste the existing Python into Windmill scripts, attach schedules, and chain them into flows with error handlers. The Community Edition runs it free on one server; Docker Compose starts three default workers at a 2 GB memory limit each.

  • An operations lead at a 15-person SaaS support team

    Give support agents buttons for refunds and account changes without direct database access

    Engineers write the scripts once; Windmill generates the forms, and an app wraps them for support staff. Those users run but do not build, so on Enterprise they count as half a seat at $10 a month.

  • A platform team letting business analysts build with AI

    Let non-engineers create automations without shipping unreviewed code to production

    AI sessions build in a forked workspace, protection rulesets block direct deploys, and an engineer reviews the diff. The model calls go to your own OpenAI, Anthropic, Bedrock or Ollama resource, billed by that provider.

Before you commit

Setup
Download docker-compose.yml, Caddyfile and .env, run docker compose up -d and open http://localhost; Helm charts cover Kubernetes. Sizing rule of thumb from the docs: one worker per vCPU and 1 to 2 GB of RAM.
Integrations
Triggers from schedules, webhooks, HTTP routes, email and WebSockets, plus Kafka, NATS and SQS on Enterprise; resource types from WindmillHub; DuckDB and DuckLake for data; an MCP server, a CLI and Python, TypeScript, Go and Rust clients. AI providers include OpenAI, Azure, Anthropic, Mistral, DeepSeek, Gemini, Groq, OpenRouter, Together, Bedrock and any OpenAI-compatible endpoint.
Data and privacy
Self-hosted state lives in your Postgres, with one encryption key per workspace for secrets; Windmill recommends encrypting the database too. Change admin@windmill.dev / changeme and the Postgres password at once, add TLS in the Caddyfile, and keep the Docker socket unmounted. A local Ollama server needs ALLOW_PRIVATE_AI_BASE_URLS. The SOC 2 report is offered to Enterprise customers.
Cost at scale
Community Edition is free beyond servers and model bills. Enterprise is priced per active seat and compute unit, with $20 per builder, $10 per non-builder and $50 per 2 GB worker in the default quote; Pro is limited to individuals, seed-stage startups and companies under 10 staff and $250k revenue.
Limits
895 open issues and pull requests. Community Edition caps workspaces at 3, users at 50 and email triggers at 100 a day; Oracle and MS SQL, multiplayer editing and Windows workers are Enterprise only.

What’s good

  • Scripts in 20+ languages become forms, endpoints and schedules automatically
  • Workflows as code, Git sync and review-before-deploy for AI-built changes
  • No execution metering on self-hosted plans
  • Bring-your-own model across OpenAI, Anthropic, Bedrock, Gemini and any OpenAI-compatible server

What’s not

  • Licence mixes AGPLv3, Apache 2.0 and proprietary code; Community images are not plain open source
  • Free edition caps SSO at 10 users and Git sync at 2
  • Default credentials, privileged workers and plain HTTP in the quickstart
  • Not accepting outside contributions beyond small fixes

Best for

Engineering and platform teams that want scripts, scheduled jobs, workflows and internal tools on one self-hosted engine with Git and code review.

Windmill alternatives

Windmill: common questions

Is Windmill worth it?
A fast, code-first workflow engine that suits engineering teams well, provided someone reads the licence and hardens the Docker defaults before it goes live. We rate it 4.3 out of 5.
How much does Windmill cost?
Self-hosted Community Edition free (unlimited executions; 50 users, 3 workspaces, SSO for 10 users). Cloud Team $10/seat/month. Enterprise from $120/month: $20 per builder, $10 per non-builder, $50 per 2 GB worker; 40% off for non-profits and universities.
What are the downsides of Windmill?
Licence mixes AGPLv3, Apache 2.0 and proprietary code; Community images are not plain open source; free edition caps SSO at 10 users and Git sync at 2; default credentials, privileged workers and plain HTTP in the quickstart; not accepting outside contributions beyond small fixes.
Who is Windmill best for?
Engineering and platform teams that want scripts, scheduled jobs, workflows and internal tools on one self-hosted engine with Git and code review.
How much does Windmill cost for a team or in production?
Community Edition is free beyond servers and model bills. Enterprise is priced per active seat and compute unit, with $20 per builder, $10 per non-builder and $50 per 2 GB worker in the default quote; Pro is limited to individuals, seed-stage startups and companies under 10 staff and $250k revenue.
Is Windmill safe for company data?
Self-hosted state lives in your Postgres, with one encryption key per workspace for secrets; Windmill recommends encrypting the database too. Change admin@windmill.dev / changeme and the Postgres password at once, add TLS in the Caddyfile, and keep the Docker socket unmounted. A local Ollama server needs ALLOW_PRIVATE_AI_BASE_URLS. The SOC 2 report is offered to Enterprise customers.
What are the best Windmill alternatives?
The closest alternatives we have reviewed are n8n, Temporal, Activepieces and Dify.

Answers checked Nov 18, 2026.

Make Windmill? Show this rating on your siteWindmill rated 4.3 out of 5 on ReviewRamble