Open SEO
every-app/open-seoReviewed by AI
Checked by RepoPicks Radar on
What is Open SEO?
OpenSEO does the everyday jobs of a paid SEO suite: find keywords, follow where your pages rank, look at competitors and backlinks, audit a site and check how AI assistants mention your brand. It runs on your own server with Docker, or on your own Cloudflare account, and there is a hosted version on the author's site. It also works as an MCP server, so Claude Code, ChatGPT or Codex can run the SEO work for you. The numbers come from DataForSEO, a paid service you sign up for and pay per request.
How easy is it to install Open SEO?
B · installer plus one simple step · Web
- Windows D
- macOS D
- Linux D
- Web B
The simplest route is the web: open openseo.so, make an account and, to get your own data, add a DataForSEO key. Self-hosting on Windows, Mac or Linux means Docker: copy .env.example to .env, fill in keys and run Compose, or use the Cloudflare route which needs a Cloudflare account and a deploy command in the terminal.
Open SEO on GitHub
- Stars
- 27,179on Nov 10, 2026
- Downloads
- 268107 this week
- Growth
- +807stars in 6 days
Stars over time
Facts
- Runs on
- Windows, macOS, Linux, Web
- You need to install
- Docker
- To build from source
- Node.js, Docker
- Disk space
- 100 MB–500 MB · estimate
- Accounts or keys
- DataForSEO (the paid source of all the SEO numbers), OpenRouter (only for SAM, the in-app AI assistant), Google OAuth, for the Google Search Console connection, Google Maps (only for the map in Local SEO), Context.dev (only to let site audits load pages that need JavaScript)
- Cost
- Free, API costs extra · The software is free (MIT). The SEO data is not: you need a DataForSEO account and pay them per request. The author's hosted version is free to try, $10/month to support the project, and the README says it charges "28% extra for every request we make to DataForSEO".
- License
- MIT (Permissive)
- Can replace
- Semrush, Ahrefs, Pro, SEO Spider, SEO
- Last change on GitHub
- Nov 7, 2026
Open SEO risk signals (5)
- Attentionusage data sent by default
compose.yaml carries "Anonymous usage heartbeat controls" with "OPENSEO_TELEMETRY_DISABLED" and "DO_NOT_TRACK" left empty by default, and package.json depends on "posthog-js" and "posthog-node". So the self-hosted copy reports usage until you set those variables.
AI review
- Attentiondocker setup runs with no sign in
The compose.yaml in the repository sets "AUTH_MODE=local_noauth", described in .env.example as "local trusted mode with injected admin user (admin@localhost)". It is only protected because the port is published as "127.0.0.1:${PORT}"; anyone who opens that port to the network, or sets ALLOWED_HOST, exposes the whole tool and its paid key without a password.
AI review
- Attentioncost lands on your card per request
README: "Bring your own DataForSEO API key and pay only for what you use." The repository even ships cost-profiling scripts ("billing:backlinks", "billing:brand-lookup", "billing:usage"), which shows single reports can be expensive; an AI agent running these tools in a loop spends your money without asking.
AI review
- Infoaffiliate link for the paid data service
README Costs section links DataForSEO as "https://dataforseo.com/?aff=255379" — an affiliate code, so the author is paid when you sign up for the service his tool requires.
AI review
- Infocrawler identity replay in site audits
.env.example: "CRAWLER_ACCESS_SECRET ... lets site audits replay a saved Shopify crawler signature", alongside the "web-bot-auth" and "http-message-sig" dependencies. The audit can therefore present a signed crawler identity when reading pages, which matters if you audit a site that is not yours.
AI review
Evidence is shown in its original language.
What Radar could not check about Open SEO
- What a month of real use costs at DataForSEO, which depends entirely on how many reports you run.
- Whether the free tier of the hosted version is enough to judge the tool.
- What exactly the anonymous usage heartbeat sends.
- Whether the Cloudflare free plan is really enough for the no-server route, as the README claims.
- How well the MCP tools behave in an AI agent, including how much data and money a single conversation can consume.