Your agent does the SEO.We give it everything it needs.
Plug VarynForge into Claude, ChatGPT, or Cursor and your agent stops guessing. We find the SEO opportunities worth writing, forge writer-ready briefs, and track every article from plan to publish.
No card required. Project creation, niche analysis, and sitemap mapping are always free.
Generic SEO tools dump data.
You wanted a plan.
The incumbent platforms ship spreadsheets full of keywords scored against the entire internet. None of them know what your site already covers. None of them tell you what to write next.
And a bare chat window is a brainstorming session, not a research engine. Your agent is only as good as the tools it can call.
VarynForge starts with your site, anchors to your niche, and forges a prioritised plan with briefs ready to ship. Then it does it again next month, with your curation intact.
Generate the plan
from Claude.
VarynForge ships a Model Context Protocol server. OAuth-authenticated, dynamic client registration, thirteen agent-callable tools. Connect Claude Desktop, Claude Code, Cursor, ChatGPT apps, or your own orchestration and run the whole workflow without ever opening the dashboard.
List projects, start research runs, pull ranked opportunities, generate briefs in batch, update the Kanban. Build a monthly report as a single Claude prompt. Pair the cockpit with whatever orchestration you already have.
See us
understand your business
before you pay.
Most SEO tools want a credit card before they show you anything. We want you to see the system read your site, profile your market, and confirm we got it right — first.
Create projects without a card
Drop in a URL or describe your niche in a sentence. We hand you a project ID near-instantly and start the analysis in the background.
Confirm we read your business right
See our reading of your market — target customer, core problem, related terms, language, country. Edit anything. Refine the strategy in plain English. No credit spent.
See your sitemap, mapped
We pull robots.txt, the sitemap index, every gzipped child sitemap. We read each page. We tell you what you already cover. Free, on every project.
Take home a customised prompt pack
Bonus on the way: download a prompt pack tuned to your niche and asset map. Drop it into Claude or ChatGPT and use VarynForge-grade context anywhere.
Six things
no one else ships.
We could list 119 features. Six of them are the moat. The rest are the proof.
Niche-centric, not URL-centric
Every report anchors to your market — the high-end coffee grinder space, the procurement-software niche, your B2B vertical. Not a list of URLs, not a global keyword cosmos. The abstraction you actually think in is a first-class citizen.
We read your site first. For free.
On project creation we crawl your sitemap, parse the pages, and reverse-engineer what you already cover. Recommendations land additive to what you have shipped — never duplicative. No credit spent.
Coverage-aware clustering
Every keyword is labelled covered or uncovered against your real content. Then we cluster the two pools separately. You stop drafting articles you have already shipped.
Briefs your writers can ship from
Outline, primary and secondary keywords, intent, internal links classified existing-vs-desired, FAQs, reference articles from real SERP data, acceptance criteria. PDF for humans, Markdown for AI agents.
Drive it from your AI assistant
Our MCP server exposes the workspace to Claude Desktop, Claude Code, Cursor, ChatGPT apps, and your own scripts — thirteen tools. List projects, generate briefs, update Kanban — without opening the dashboard.
Curation that survives every re-run
Mark a competitor as important. Ignore the noise domains. Pin the queries that matter. Re-run research next month — your judgment is preserved, not overwritten.
A Kanban of articles,
briefed and queued.
Five stages — Planned, Generating Brief, Brief Ready, Out for Writing, Monitoring. Each card carries a writer-ready brief. Download triggers the status change. Your roadmap is auditable, defensible, and always one click from execution.
Each run keeps your radar
lit for thirty days.
Complete a research run and the radar switches on: rising narratives in your niche, synthesized daily from news, Reddit, and Hacker News discourse. Hot picks arrive flagged act-within-48h, and one keep sends the pre-picked articles straight into your content plan.
Clear the sweep from your phone with your first coffee — swipe right to forge, left to pass — or let your MCP-connected agents watch the radar and file the article suggestions before you sit down.
We will not pretend
we do everything.
We do not crawl backlinks. We do not poll Google to update your positions every week. We do not audit Core Web Vitals. We do not run in twelve languages.
We do watch your market. Every research run keeps a radar lit for thirty days — rising narratives in your niche, synthesized daily from news, Reddit, and Hacker News discourse. That is emergent-narrative radar: what your market is starting to talk about. It is not rank tracking, and it is not a Google Alert.
We do one job. We forge a prioritised content plan with briefs, anchored to your niche and your site.
If you need link-graph dashboards or daily position polling, you have other tools for that. We will tell you straight.
And we are built agent-first. Every surface works from the dashboard too — but if the words MCP, Claude Code, and Cursor mean nothing to your workflow yet, we are probably ahead of where you are. Come back when you hire your first agent.
Start with the part
that is free.
Drop in a URL. Watch us read your site. Confirm we got your niche right. Then — and only then — decide whether to forge.
Start a Free Project