OpenSEO review: open-source Semrush alternative, first look

An SEO tool trending on GitHub is rarer than another agent framework. When this one hit my feed twice last week, I read it. OpenSEO calls itself an open-source alternative to Semrush and Ahrefs; GitHub counted 18,213 stars as of September 10, 2026. I have not run it on a live project yet. Treat this as based on the README, the pricing page and the docs. Half of what gets written about such tools is launch-post copy-paste.

What OpenSEO runs on, and why that part matters

The load-bearing fact that most launch coverage skips: OpenSEO is a wrapper, and the data underneath is DataForSEO. The platform itself is TypeScript and MIT-licensed, with 2,300 forks. It covers six workflows: keyword research, rank tracking, competitor insights, backlinks, site audits, and AI visibility. Every workflow that touches search data calls DataForSEO’s API, which means you pay DataForSEO’s rates either way. Open source describes the code you can inspect and modify, which is not the same as free data.

The second structural fact is the agent layer. OpenSEO ships an MCP server, which lets Claude Code, Codex and similar clients call the same tools your browser can. The tool list covers keyword research, SERP inspection, backlink overviews, rank tracking, and even Google Search Console performance. A plugin install adds nine Agent Skills on top: prebuilt workflows for common SEO tasks. No incumbent suite offers that natively, which is why the repo trends with builders rather than SEOs. The MCP setup docs walk through the client wiring for Claude Code, Claude Desktop, and Cursor.

Self-hosting is documented on two paths, and the README is honest about which is which. Docker is the “best for testing it out” option on your own machine, while Cloudflare is the recommended path. Cloudflare is recommended for anything internet-facing or shared with a team, and it works on the free plan. Either way you bring your own DataForSEO key, which is the real setup cost. The tool is free; the data pipeline is not.

The pricing math to run before you sign up

The hosted plan is $10 per month, and that price includes $10 of usage credit. The floor is genuinely $10 unless you outgrow it. The pricing page’s own estimator prices a plausible solo month at $7.12 of usage. The vendor sketches that month as 100 keyword searches, 20 backlink checks, and weekly rank tracking on 50 keywords. That fits the included credit with $2.88 of headroom. This is the vendor’s scenario rather than mine, though the arithmetic is checkable line by line.

The per-request prices are the useful part, because you can project your own month from them. Keyword searches run about $0.05 each at typical result limits. A domain overview with a year of history runs about $0.08. Rank tracking is cheap at roughly $0.54 for 217 checks, while AI visibility is where budgets go to die. The expensive line item is AI visibility: a ChatGPT brand check runs about $1.09. A daily brand-monitoring habit costs more than the plan itself. Tracking AI citations is its own problem, which I covered in llms.txt and what actually moves AI citations.

Two structural details about the pricing deserve attention. Self-hosting runs slightly cheaper, because the hosted service charges a 28% markup on every DataForSEO request. When your credits run out, the app shows errors rather than charging you silently; the FAQ states this plainly. Top-up credits never expire. There is a free trial with $0.50 of credits, enough to verify the thing works before you pay.

For comparison, the same pricing page quotes Ahrefs’ cheapest plan at $129 per month. I could not verify that number directly, because ahrefs.com serves regional pricing. The page I loaded from Germany quotes Lite at €119 per month as of September 2026. That lands near $129-140 on the day’s rate, so call the entry tier €119/$129 per month, roughly 13x OpenSEO’s floor. Its data index is far deeper, though. You are not buying the same thing at either price.

Where it fits a marketing stack, and where it does not

OpenSEO fits as the cheap, controllable floor: keyword research, rank tracking, and agent-run reporting. For one person or a small team, that lands at roughly a tenth of suite pricing. It also fits a specific builder pattern. The MCP layer lets your agent pull SERPs and backlink data inside the session where it writes the brief. That pairs naturally with the AI briefs workflow: research and writing end up in one place. For teams running programmatic SEO with discipline, per-request credits keep costs legible because no seat fee buries them.

It does not fit everywhere, and the gaps are structural. DataForSEO’s backlink index is not Ahrefs’ index, so backlink analysis at client scale still favors the incumbent. The rank tracker is young, and the audit feature stays at page-level signals. There is no deep crawling, no local rank grid, and no content optimization module. It is a floor rather than a suite. Agencies that live in white-label reports will not move off Semrush yet, because the reporting depth is not there. If your month looks like 5,000 keyword searches across 30 projects, run the credits math first. The discipline is the one I applied to ChatGPT for marketing. Separate what the docs promise from what you have watched work.

• You run SEO for one site or a small portfolio, not a 30-client roster.
• Your core loop is keyword research, rank tracking and audits, not deep backlink forensics.
• You want an agent calling your SEO tools, and you accept it spends your credits.
• You can tolerate a v0.1.x product: releases move weekly and things break.
• You already have, or can create, a DataForSEO account.

How to read 17k stars on a tool like this

The growth is real and verifiable, and the shape of it names the adopters. Star velocity alone never tells you which crowd showed up. The repo passed 1.6k stars at the end of April 2026, per its Show HN post. By September 10, the GitHub API showed 18,213. That is ten-fold growth in under five months, but what interests me more is the channel pattern. Hacker News barely engaged, with the April post at 2 points and an August resubmission at 1. GitHub trending did the carrying. A tool the HN crowd scrolled past is getting adopted by SEO operators, which matches who the product serves.

The activity signals look healthy for a young project, since releases shipped on a regular clip through the summer. The tags ran v0.1.3 on July 30, v0.1.4 on August 9, and v0.1.7 on September 2. Two more releases landed in mid-August. The repo shows 158 open issues and 56 watchers. The watcher count is the honest usage proxy, since watching costs more intent than starring. Commit messages through early September show bug fixes on GA4 date ranges and audit retry logic. That reads as a team using its own tool and fixing what squeaks. The Adobe-Semrush acquisition from November 2025 hangs over all of this. Adobe paid $1.9 billion to consolidate the category, and the open-source floor showed up within a year.

Filed under SEO & Content, alongside the programmatic SEO workflow, the AI briefs workflow, and the llms.txt plumbing post.