DocuWriter SEO Audit: Plugins, Public Docs, Not MCP Pages

DocuWriter.ai generates structured docs from a connected repo: architecture, APIs, diagrams, READMEs, kept in sync with Autopilot. Domain rating is fine. The organic story is not.

DR 45 is a usable start. Traffic and impressions are down. Pages were cut. AI Overviews and crawler changes ate visibility. The remaining organic is not the product pitch. The cheat sheet is 733 visits — about 72% of the site.

The Python cheat sheet is the site

The page that grew is a Python cheat sheet: operators, conditions, copy buttons. A Try DocuWriter button sits on it. What DocuWriter actually is only appears after you click through. That is the mismatch. The keyword that pays is a list of syntax, not “complete structured documentation for entire codebases.”

The homepage repeats that long pitch twice. Developers skip it. Most of them already have an agent, a skill, or a hook that writes docs.

Pricing and the MCP dead end

The product is priced like a team platform. The FinLocker case (100 repos in five weeks) only makes sense on Unlimited. MCP is listed as the integration story. That is the wrong install path.

People in Cursor, Claude Code, and Codex do not want to wire an MCP server. They want a plugin they can install, or an agent that installs it. Package a Codex plugin and get listed.

Cursor has a plugin marketplace and an MCP registry. A pull request to get listed beats another “connect MCP” docs page. Discoverability lives in the IDE, not in another SEO article.

“AI code documentation” is not a category anymore

Brand search for DocuWriter is small and falling. AI code documentation used to rank. Now the SERP is OpenAI, OpenCode, Reddit, IBM, CodeGPT, Medium, index.dev, and a 22-star GitHub repo. DocuWriter is not in the top 10. Sometimes an AI Overview. Sometimes an image pack.

Reddit comments on a three-year-old thread are cheap. Paying developers for this keyword is not. Search volume died because IDEs generate /docs, Notion, or an MCP on demand. Listicles for “AI documentation tools for legacy code” pull ~25 visits. index.dev is at zero. Do not keep building the same keyword.

The blog lost clicks, then rankings

Posts like 7 UML examples look generated: AI images, unhighlighted code, thin examples. Other posts were deleted. The technical documentation templates roundup has screenshots for competitors and none for DocuWriter. Developers spot the slop, bounce, and the ranking decays. Keep the domain if it still makes money. Rewrite the posts as real information, or stop publishing filler.

Mintlify is a different job. Open source is free.

Mintlify is big because it is a public docs site: API reference, quick starts, Markdown as a product website. Internal docs belong in the repo.

DocGen is open source with a Claude Code skill. autocodedocs.ai feels a year late. Developers will not pay for what a free skill already does.

The trick: host public docs like CodeWiki

CodeWiki is the pattern to steal. Host generated docs for public open-source repos. Programmatic pages like /flutter/flutter. Show the output, then charge for private repos. Same split as GitHub: public free, private paid.

No free plan means no library of show-don’t-tell pages. Add one. Let AI and Google find “how to set up React” style pages on your domain. You still need links. W3Schools is old. CodeWiki will rank those deep pages. Fill missing docs for older libraries. That is the SEO product.

TL;DR

Ship Codex and Cursor plugins. Stop selling MCP setup as the growth loop. The Python cheat sheet is the organic site; the homepage is too long. “AI code documentation” is a dead SERP. Rewrite or kill slop posts. Host public repo docs like CodeWiki, free for open source, paid for private. Ask whether the SaaS is still worth the rewrite.