Back

SEO Agent

Find keyword demand, sync Google Search Console, and ship page-level fixes — a Growth Autopilot sub-workflow reviewed in minutes.

The SEO Agent is a sub-workflow inside Growth Autopilot that takes you from raw keyword demand to applied page fixes without leaving your normal review loop. You connect Google Search Console once, and the agent handles scheduled data collection, keyword mining, and writing per-page recommendations — you spend a few minutes approving or dismissing them.

How the Pipeline Works

The SEO Agent runs in four stages:

  1. Keyword mining. The agent researches what your target audience is searching for, cross-references competitor keyword gaps, and adds candidates to your keyword list with volume and difficulty signals.
  2. Page inventory and GSC sync. A crawler indexes your landing pages, feature pages, blog posts, and docs. Google Search Console data (clicks, impressions, CTR, position) is pulled in automatically to show real performance alongside keyword gaps.
  3. Recommendations. The agent maps keywords to pages, identifies title, H1, meta, content, and internal-link opportunities, and writes a specific recommendation for each — including a copy-paste prompt you can hand to a coding agent.
  4. Review and re-evaluation. You apply or dismiss each recommendation in the dashboard or via MCP. After roughly 28 days, the agent re-evaluates applied recommendations against fresh GSC data to see whether the change moved the needle.

Two Layers Under the Hood

The SEO Agent is split into a program layer and an agent layer. The program layer handles scheduled crawls and GSC data syncs — deterministic, cheap, and always running in the background. The agent layer applies judgment: it mines keywords by reasoning about your product category, ICP, and competitor coverage, and it writes recommendations by evaluating page content against real search-performance data. Keeping these layers separate means you always have fresh facts before the agent spends tokens on analysis.

Three Ways to Use It

Dashboard. Go to /console/growth and open the SEO area. The Keywords, Pages, and Recommendations tabs give you the full picture, and the Google Search Console connection card lets you link GSC to unlock performance data.

IDE or terminal. MCP tools (list_keywords, list_seo_recommendations, update_seo_recommendation) let you query and act on your SEO data directly from Cursor, VS Code, or any terminal agent. For multi-step work — running a keyword sweep or generating a batch of recommendations — use growth_task with a natural-language request.

Automatic schedule. Keyword research and page crawls run on a weekly schedule so your data stays current without manual triggers.

Plan Access

Viewing keywords, pages, and recommendations in the dashboard and through MCP list tools is available to any signed-in user who has a growth product. The AI-driven audits — keyword research and page recommendations generated by the agent — require the Growth plan ($100/mo).

Next Steps

  • — How the agent finds and scores keyword opportunities
  • — Connecting GSC and reading page performance data
  • — Reviewing, applying, and dismissing page-level fixes
  • — Using MCP tools and growth_task for SEO work in Cursor or the terminal
  • — What's included per plan, cron schedule, and upcoming features
Back
SEO Agent | VibeCom Docs