every-app/open-seo

Data

A pay-as-you-go SEO suite with keyword research, rank tracking and audits, plus MCP access for AI agents.

24,938
3,402
TypeScript
MIT
8 months old

OpenSEO is worth a look if SEO data costs are a real line item for you and you are willing to handle your own hosting and bring a data vendor key. It is less appealing if you want a finished SaaS with included data and no operational work.

The problem

A small team or solo marketer wants to check keyword volumes, watch rankings, compare backlinks and audit a site. With conventional SEO suites, that means a recurring subscription and a large interface that bundles many features the user never opens. OpenSEO changes the middle of that workflow: instead of paying for a bundled data plan, the user runs the app and connects it to DataForSEO, paying that vendor for the queries actually made. The result is the same kind of research and tracking workspace, but the spend tracks usage and the data source is visible.

What's different

The distinguishing part is not the workflow list—keyword research, rank tracking, competitor insights, backlinks, site audits and AI visibility are conventional—but how the app is packaged. It is self-hostable, MIT-licensed, and exposes an MCP server so agents such as Claude Code, OpenClaw or Hermes can query the same SEO data through prebuilt or custom skills project instructions. That makes it possible to script repetitive SEO checks rather than clicking through a dashboard.

The trade-off is that OpenSEO supplies the interface and orchestration, not the underlying index. Every SEO number comes from DataForSEO, so the quality, coverage and cost of results are effectively that vendor's. The project also positions a hosted plan at $10/month and notes that the hosted service charges a 28% markup on DataForSEO requests pricing; self-hosting removes that markup but shifts hosting and key management onto you.

Who should try it

Good fit: a developer or technical marketer who already has (or is willing to open) a DataForSEO account, wants a focused SEO workspace, and would use the MCP endpoint to pull rankings or audit results into an agent-driven workflow. It also suits someone who prefers to read and modify the code rather than wait for a roadmap.

Poor fit: a marketing team that wants an all-in-one platform with included data, no API billing and no server to maintain. Self-hosting has two documented paths—a simple Docker setup described as best for testing on your own machine, and a recommended Cloudflare deployment for internet-facing use across devices or with a team Docker guide, Cloudflare guide. Choosing between them is itself part of the adoption decision.

Requirements and limitations

The hard dependency is a DataForSEO API key; without it there is no SEO data, and the project points to a separate setup guide API key guide. Costs are therefore metered by DataForSEO rather than flat, so heavy keyword and backlink research produces variable bills rather than a predictable subscription. Users moving from an all-inclusive suite should model query volume before assuming it is cheaper.

Operationally, self-hosting is not a one-click install. The Docker path is framed for personal testing, and the README says simpler hosting on Railway, Coolify or Dokploy is planned for the next few months—so those platforms are not a documented supported path today. The Cloudflare route works on the free plan but still requires standing up and maintaining the deployment. The agent features also depend on configuring MCP and skills, which is separate work from the web UI MCP setup, skills setup.

HotRepo's take

HotRepo's only dated observation is 2026-10-25, when the project had 24,873 stars and gained 142 in a day while appearing on the daily trending board. That is a single snapshot, not a growth curve, so persistence of interest cannot be judged. The practical check before committing is whether your expected DataForSEO query volume, plus the hosting effort, actually beats the subscription you are replacing.

Written by AI from the project's README, updated October 25, 2026. The project's own docs are the source of truth.