npx @seoskillsai/cli
Content & Entities Cursor IDE

Write SEO Articles with Cursor: Composer, Project Rules, and the seo-content Skill

DEEP CURSOR IDE TECHNICAL WORKFLOW

Human-Edited & Verified • Deterministic AI-assisted research. Editorial standards.

Engineer at a dual-monitor desk using Cursor Agent to write seo articles with cursor into a markdown file, not a chat pane AI Edited
AI-generated Composer session writing into a repo file. Raw provenance file.

Last updated: 3 September 2026

Write SEO articles with Cursor by pinning an inspectable seo-content skill, a project rule, and an HTML-first file the crawler can GET without waiting for Agent chat.

TLDR

  • First-party GSC (3 Aug–31 Aug 2026) on this URL: 2 clicks, 15 impressions, blended position ~33. The page was still a PARTIAL agent×skill template in that window.
  • Google rewards people-first pages, not mass AI filler (helpful content).
  • Cursor project rules live in .cursor/rules as .mdc files. Root .cursorrules is legacy.
  • MCP pulls live tools. It does not replace sourced claims.
  • Run npx @seoskillsai/cli add seo-content --agent cursor.

Contents

Install the seo-content skill in Cursor

The one-command install drops the skill into the agent the CLI knows as cursor. That is the Cursor recipe. The agent-agnostic SOP still lives on the content pillar.

npx @seoskillsai/cli add seo-content --agent cursor
npx @seoskillsai/cli run seo-content --agent cursor --target "https://example.com"
  1. Open the site repo in Cursor (the folder that actually deploys HTML).
  2. Run the add command above so skills/seo-content/SKILL.md is on disk.
  3. Attach the skill in Agent (@seo-content or the CLI-generated rule) before you ask for a draft.
  4. Point Agent at a real path (src/data/markdown/... or src/pages/...), not “paste a blog into chat.”

Cursor’s own catalog for this site is the Cursor hub. Sibling agents (Cline, Claude, ChatGPT) have their own /{agent}/seo-content twins. Do not retitle this URL to steal those queries.

Pin project rules so Agent writes files, not chat

Developer reviewing .cursor/rules markdown files beside an HTML-first article draft on a second screen AI Edited
AI-generated project-rules review beside an HTML draft. Raw provenance file.

Cursor documents four rule types. For ranking copy, use a project rule with globs on your content files, or an AGENTS.md in the repo root. Official docs: Cursor Rules.

A useful .mdc is short:

  • Draft into the locked URL’s file. Never invent a nicer slug.
  • Primary phrase only in title, H1, first sentence, meta, featured alt.
  • Every statistic needs a fetched URL. Refuse “studies show.”
  • First-response HTML must contain the H1 and body. No canvas-only headlines.

The content skill pillar is the agent-agnostic E-E-A-T / algorithmic-authorship SOP. This page is how Composer applies that SOP inside Cursor.

Open the Cursor hub

Surface What crawlers see When to use it
Agent chat only Nothing at the URL Scratch notes, never the ranking page
Markdown/HTML in the repo (SSG) Full headings and links on GET Default for this site
MCP tool output Only if you write it into a file Live GSC/SERP numbers, then paste with a source URL

Use MCP for live data, not invented SERP facts

Cursor connects Agent to servers through mcp.json (project .cursor/mcp.json or user ~/.cursor/mcp.json). Docs: MCP integrations.

Use MCP when you need live Search Console rows, a crawl, or a SERP API. Do not use an MCP scrape as a substitute for original reporting. Google’s people-first questions still apply: unique analysis, no preferred word count, no date-bump without a real change.

The directory of JSON-RPC adapters on this site is the MCP extensions hub. Pair a DataForSEO or Firecrawl server with the skill; keep secrets in env, not in the article.

Using AI doesn't give content any special gains. It's just content.

Google Search Central, February 2023

If the model cannot fetch a URL, it cannot cite a tweet, a study, or a “benchmark.” Leave a TODO. That is the same gate as Google’s AI-content FAQ: automation is fine when the page is helpful; it is spam when the point is ranking manipulation.

Publish HTML the crawler can read without Agent

Composer is not the index. Ship SSG or static HTML so Googlebot, Bingbot, and SeznamBot receive the H1 and body on the first response. This site’s editorial/benchmarks surface is the technical guides hub. Agent recipes like this URL sit next to those guides, not inside a JS-only feed.

Checklist before npm run build:

  • Unique title, meta description, canonical, and contiguous H1 in the initial HTML.
  • Internal links are real <a href> paths (/cursor, /skills/seo-content).
  • Images: WebP ≥1200px wide for the featured slot; visible AI badge + raw file.
  • House Byline under the H1; policy link to /skills/seo-factuality.

A “best Cursor SEO copywriting prompt” is a saved rule plus a locked file path—not a one-shot chat that never reaches dist/.

FAQ

How do I draft ranking copy inside Cursor?

Install seo-content for the cursor agent, attach a project rule, and tell Agent the file path. Review the diff. Build HTML. Do not publish from the chat pane.

Do I need MCP to draft SEO copy?

No. MCP is for live tools (GSC, crawl, SERP APIs). Rules plus the skill are enough for structure. Add MCP when a number must come from an API, not from memory.

Cursor vs ChatGPT for this workflow?

Cursor Agent edits the repo. ChatGPT Actions live in a different runtime. Use /chatgpt/seo-content for that twin. This URL stays the Cursor recipe.

Where is the agent-agnostic content skill?

/skills/seo-content. Cline’s entity-grounded twin is /cline/seo-content.

Will a crawler see a Composer-only draft?

Only after you write it to a public HTML URL. Chat history is not a document the index can GET.

Related reading

Sources

~6,400 Avg. Token Consumption
22s Avg. Execution Latency
$0.022 Estimated API Cost / Run
100% .cursorrules Active
RUNTIME CONFIGURATION

Cursor IDE Configuration Blueprint (.cursorrules)

Paste this configuration block directly into your .cursorrules:

.cursorrules JSON 
{
  "skills": {
    "seo-content": {
      "antiFillerGuard": true,
      "minInformationGain": 0.85,
      "targetHeadingTree": "strict"
    }
  }
}
SEMANTIC ENTITY-ATTRIBUTE MODEL

What the Cursor .cursorrules Semantic AI Copywriting & Entity Grounding Analyzes

Automated diagnostic data points evaluated during every execution run.

Diagnostic Category Specific Data Points Checked Algorithmic Impact
Information Gain Score Original statistics, proprietary code examples, unique teardowns Exempts content from Google unhelpful content suppression
Entity Salience & Triples Subject-Predicate-Object grammars in first 100 words Triggers instant AI Overview snippet extraction
STEP-BY-STEP WORKFLOW

How to Execute Semantic AI Copywriting & Entity Grounding in Cursor IDE

Imperative configuration instructions with ready-to-run commands.

1

Step 1: Extract SERP Triples

Scrape top 10 search results to find unaddressed competitor gaps.

seoskillsai content research "claude seo mcp"
2

Step 2: Generate High-Salience Content

Author full semantic articles adhering to strict heading hierarchies.

seoskillsai content generate --template skill-article
MACRO PILLAR GUIDE

Looking for universal benchmarks and cross-agent comparison?

Read the comprehensive universal guide for Semantic AI Copywriting & Entity Grounding.

View Universal Pillar Guide →

Make SEO Skills AI a Preferred Source

Keep SEO skill docs highlighted in Google Search and AI Overviews.

Add on Google
Preferred Source added