Your agent runs the ranking loop. You approve at the gate.
Connect Kitecraft to your own Claude or Cursor agent over MCP. It baselines you across ChatGPT, Gemini, Google AI Mode, Google AI Overviews, and Perplexity, builds the 90-day plan, drafts the content, and queues it to ship — seeded by your context. Money and publish are the only human gates.
Works with your existing agent · Two human gates · Full audit log
Autonomous, not unsupervised
One MCP connection. The whole loop.
Kitecraft already knows what to track — it turns your real Google search demand into the AI-answer queries worth monitoring. Your agent drives the rest.
The moat is auditable autonomy
Autonomy you can trust is autonomy you can audit. Every paid scan and every publish passes through the same control plane — and this substrate already exists.
- Policy-checked: every action is evaluated against your rules before it runs.
- Approval-gated: spending money and publishing content are the only steps that require a human — you.
- Rate-limited: scans and runs respect hard ceilings, so an agent can't run up your bill.
- Audit-logged: every scan, draft, and publish is recorded with who, what, and when.
Everything between baseline and publish runs on its own. You step in twice.
Money
Paid scans wait for your confirmation. You decide what coverage is worth running before a cent is spent.
Publish
Drafts wait in your queue. You read, edit, and approve — then it ships. Kitecraft does not auto-publish behind your back.
Do I need a Kitecraft-specific agent?
No. Bring your own Claude or Cursor agent. It drives Kitecraft over MCP, seeded by your context — no new tool to learn.
Can the agent spend my money or publish on its own?
No. Paid scans and publishing are hard-gated to your explicit approval. The agent can do everything up to those two lines and then it waits for you.
Do you guarantee rankings or citations?
No. Kitecraft baselines, plans, and ships, then re-checks the same queries so you can see observed movement. Rank Score reports what actually changed — it does not promise a result.
What is the audit log for?
Every paid scan, draft, and publish is recorded. If you ever need to answer 'why did this run' or 'who shipped that,' the log already has it.
Run the loop
Start the trial, connect Google, and point your agent at Kitecraft over MCP. Baseline, plan, draft, ship, re-check — with money and publish under your control the whole way.