OpenSEO alternative: Serpkin vs OpenSEO for coding agents

OpenSEO and Serpkin both plug into Claude Code, Codex and Cursor over MCP, but they do different jobs. OpenSEO is an open-source SEO data layer: keyword research, SERPs, backlinks, rank tracking, site audits, local SEO and AI visibility, plus skills that tell your agent how to use that data and write a report. Serpkin turns your site and search data into ranked missions with a patch-ready plan, then checks the change on your preview and on production and measures the result at 14 and 28 days. If you want cheap, broad, self-hostable data and you are happy to decide what to change, pick OpenSEO. If you want your agent to get one concrete job at a time and someone to check it got done, pick Serpkin.

We make Serpkin, so weigh that. Every OpenSEO fact on this page comes from openseo.so or its GitHub repository, checked on November 5, 2026 and listed under sources. Where we couldn't confirm something on their site, we left it out.

#What OpenSEO is

OpenSEO's GitHub README calls it an "open source alternative to Semrush and Ahrefs". The code is MIT licensed and lives at every-app/open-seo, which had more than 26,000 stars when we checked. Its homepage says it is "100% open source", and it names its third-party data provider on the homepage. You can run it in two ways:

• Hosted at app.openseo.so, with an MCP server at https://app.openseo.so/mcp and plans billed by usage.
• Self-hosted with Docker or on Cloudflare, per the README. You need an API key from that data provider, and OpenSEO's "Why OpenSEO" page says self-hosted users "cover their own hosting and data costs."

The feature pages and MCP docs list these tool areas:

• Keyword research with volume, difficulty, intent, CPC and live Google SERPs, plus saved keywords.
• Domain overview, ranking keywords for a domain or page, and backlink and referring-domain data.
• Rank tracking on desktop and mobile, with rank history.
• A site audit that "crawls, renders and sorts every issue by severity."
• Local SEO: local business search, Maps and Local Finder SERPs, Google Business Profile audits, reviews and a Maps grid rank check.
• AI visibility: brand mentions and citations in ChatGPT, Gemini and Google AI Overviews, and a prompt explorer.
• Google Search Console performance and URL inspection, plus IndexNow notifications to Bing.
• Shared project context (business, goals, competitors, key pages) and saved HTML reports.

On top of the MCP server, OpenSEO ships skills: SKILL.md files for jobs such as SEO project setup, SEO audit, keyword research and clustering, competitor analysis, local SEO, AI visibility audit and link prospecting. Each skill delivers its findings as an HTML report saved to your project. There is also a Strategy Library of SEO playbooks grouped by topic. In Claude Code and Codex CLI, the docs recommend the OpenSEO plugin, which installs the MCP server and the skills in one step. If you only want the MCP server:

claude mcp add --transport http --scope user openseo https://app.openseo.so/mcp
# Codex CLI
codex mcp add openseo --url https://app.openseo.so/mcp

OpenSEO also markets a Google Search Console MCP with two read-only tools (search performance with up to 16 months of history, and URL inspection for up to 10 URLs per call). On hosted OpenSEO it needs an account and the $10 plan, and the Search Console tools use no credits.

#What Serpkin is

Serpkin reads your site, your Search Console data once connected and its own research, then ranks the SEO work into missions. A mission is one task for one site, with target pages, ordered steps mapped to your source files, code snippets where it can generate them, and acceptance checks. Your agent gets it over MCP (32 tools) or the CLI. You set it up from your repository:

npx serpkin            # connect Claude Code, Codex or Cursor and register the site
npx serpkin check https://my-branch.vercel.app   # compare a preview with production

After the agent ships, Serpkin verifies the mission on the preview and again on production, and a deploy hook can trigger the production check for you. Then it takes a baseline and measures at 14 and 28 days, comparing the pages you changed with similar pages you didn't. Around that loop sit rank tracking, site watch, competitor watch, a topic map and research plan, and a weekly summary by email, Slack or Discord.

#The core difference: data versus a checked job

Both tools give the agent context. The difference is what happens after the agent reads it. With OpenSEO, the output of a session is research: a keyword list, a competitor breakdown, an audit report. What to change in your code, and whether it worked, is up to you and your agent. OpenSEO's MCP docs describe write actions inside OpenSEO (saving keywords, creating rank trackers, updating project context, saving reports, IndexNow pings), but no tools that turn a finding into a code change or check one on your site. Its roadmap lists scheduled site audits, rank tracking alerts and daily or weekly email reports as "Soon", not shipped.

With Serpkin, the output is a specific change and a record of whether it landed. Neither tool edits or deploys your site itself: in both cases your coding agent writes the code and you merge.

#Feature comparison

#Pricing compared

How OpenSEO bills

Every OpenSEO plan includes a monthly amount of usage, and each action costs "what the data costs us plus a flat 28%", rounded up to the nearest tenth of a cent. The pricing page gives examples: about $0.05 for a keyword search and up to 1.5 cents per keyword for a rank check. Projects, saved results, exports and Search Console data don't use credits. Included usage resets each billing cycle; top-ups roll over. The free trial is $0.50 of credits, and you need a subscription for rank checks and prompt research.

For light research on one site, that is cheap. The pricing page's own estimate for 100 keyword searches, 20 backlink checks and 50 keywords tracked weekly comes to about $10 a month. Heavy research scales linearly, which is fair but means your bill follows your agent's curiosity, so set the auto-reload limit.

How Serpkin bills

Serpkin charges a flat plan. Builder is $29 a month or $290 a year, Pro $79 a month or $790 a year, and both start with a 7-day free trial (card required). Credits only cover research: a keyword research run uses 2, and Builder includes 500 a month, Pro 1,500. Scans, foundation missions, result checks, rank tracking, competitor watch and content reviews use none. You pay more than OpenSEO's entry plan, and what you pay for is the mission loop rather than raw lookups.

#Where OpenSEO is the better pick

• You want open source or self-hosting. MIT license, Docker or Cloudflare, your own data account. You can read and change every tool. Serpkin is a hosted service.
• You want the lowest entry price. $10 a month with $10 of usage included, or self-hosted at cost. Serpkin starts at $29.
• You need broad raw data. Domain overviews, backlink profiles, live SERPs, competitor SERP comparisons and a prompt explorer across AI models. That is closer to a lightweight Ahrefs or Semrush than to Serpkin.
• You do local SEO. Google Business Profile audits, reviews and Maps grid rank checks are in the MCP server today. Serpkin has nothing comparable.
• You track AI answers. OpenSEO tracks brand mentions and citations in ChatGPT, Gemini and AI Overviews. Serpkin checks whether AI crawlers can reach your site, not who gets cited.
• You are an SEO or agency doing research for clients. Saved reports, project context and an Agency plan with hands-on agent setup fit that work.

#Where Serpkin is the better pick

• You want the agent to get a job, not a report. serpkin_next_mission returns one ranked task with target pages, files, steps and done-when checks. You don't have to turn research into a plan.
• You want proof the change landed. The preview check fails on regressions (a new noindex meta tag, a missing title or canonical, a broken sitemap) before you merge, and the mission is verified again on production.
• You want to know what each fix did. Results at 14 and 28 days, net of similar pages you didn't change, labeled as correlation. OpenSEO gives you the data to do this comparison yourself.
• You want it to run without you. Deploy hooks, site watch, competitor watch, a weekly summary and an optional weekly pull request from a GitHub Action.
• You ship sites, not SEO research. If you are a developer or founder who wants SEO done between features, a short list of checked tasks is less work than reading reports.

#Using both

They don't conflict. A setup some teams will like: OpenSEO for research (local, backlinks, AI citations) and Serpkin for the fix, check and measure loop. Add one MCP server at a time so you can tell which one the agent is calling, and tell the agent which to use for what in your AGENTS.md or CLAUDE.md:

## SEO tools
- Research questions (keyword ideas, backlinks, local, AI citations): use the openseo MCP server.
- What to change next, and checking a change: use serpkin (serpkin_next_mission, npx serpkin check).
- Ask before any call that costs credits on either tool.

If you only run one, the question is simple: do you need more data, or do you need the work done and checked?

#Which to pick

To try the Serpkin side, run npx serpkin in your repository, or check one URL first with the free SEO checker (no sign-up). Setup and the mission loop are in getting started, and the other servers worth knowing are in SEO MCP servers compared. Both Serpkin plans start with a 7-day free trial. Serpkin doesn't merge, deploy or promise rankings.