DEV Community

Perufitlife
Perufitlife

Posted on

I shipped 5 things around my product in 90 minutes — MCP server, GitHub Action, 3 SEO landings

#indiehackers #mcp #githubactions #seo

I shipped aitells (free AI text detector + paid humanizer) yesterday. Today I shipped 5 more things around it in about 90 minutes. Sharing because each is a small, reusable distribution play that compounds.

1. MCP server (npm published, GitHub repo)

→ @perufitlife/aitells-mcp · github.com/Perufitlife/aitells-mcp

Claude Code, Cursor, and any MCP-compatible client now get detect_ai_tells and humanize_text as native tools. Configure once, ask the model "humanize this LinkedIn post in my voice" and it picks the right tool.

Implementation is 100 lines of TypeScript wrapping the existing aitells public API. The win isn't the code — it's that the MCP ecosystem has its own discovery channels (Glama.ai, awesome-mcp-servers, etc.) and a different audience than the web app.

{
  "mcpServers": {
    "aitells": {
      "command": "npx",
      "args": ["@perufitlife/aitells-mcp"]
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

2. GitHub Action (marketplace listing)

→ github.com/Perufitlife/aitells-action

Scans PR titles, bodies, and commit messages for em-dashes, "delve", parallel bullets, etc. Posts a friendly summary comment on every PR.

- uses: Perufitlife/aitells-action@v1
  with:
    target: both
    fail-on: 65    # block merge if humanness < 65
    comment: true
Enter fullscreen mode Exit fullscreen mode

Same idea as the MCP server — wrap the existing API in a different package format to hit a different audience (engineering teams that care about keeping their PR history sounding human).

3-5. Three competitor SEO landings

Each one is a Next.js page with metadata + canonical URL + a shared _vs/client.tsx component that does live detection on whatever text the user pastes. The differentiation argument is in the page copy. Templated, took ~30 min total for all three.

People searching "zerogpt vs..." have buyer intent. That search has high commercial value for low effort.

The meta pattern

I had one product (aitells.vercel.app). I now have 6 entry points to it: the web app, MCP server, GitHub Action, and 3 SEO landings. Plus the existing Dev.to articles and GitHub profile README and cross-links from other repos.

The actual code under all of these is the same /api/detect and /api/rewrite endpoints. Everything else is packaging the same core for a different audience.

Building takes hours. Packaging takes 30 minutes per format. If your product solves a real problem (and AI text leaking through review pipelines is a real problem), the bottleneck isn't building more features — it's getting the existing thing in front of the audiences that don't know your URL.

I also just published the public API docs so anyone can integrate the raw endpoints in their own product. The detector is free forever. The humanizer is $19 lifetime, first 100 buyers only, then $49/mo.

If you want to use it as an MCP server: npm install -g @perufitlife/aitells-mcp and follow the config snippet above.

I also build supabase-security and sister BaaS auditors. Same pattern: build the tool, then package it 6 ways. The pattern works.

Top comments (0)