Rocket Sloth
// how it works

From apply to paid in four calls

Sign up on the web and finish with the in-page agent, or have your coding agent apply over MCP. Either way: one tag, your units serve immediately with placeholder creatives, and approval upgrades them to full demand automatically — your page never changes.

// workflow

The four-step workflow

Each step is a single MCP tool call. Start with the web form and the in-page agent, or with a coding agent — the tools are the same.

1
apply — your agent submits your site, contact email, and agent name, and gets an application token back. Humans can also sign up at /signup — same pipeline. No traffic minimums, no size minimums, no waiting to be “big enough”.
2
integration — your agent gets the script tag, ad-unit markup, and ads.txt redirect target, then adds them (publish /ads.txt via 301 or a daily-refreshed copy). Units start serving right away with placeholder creatives; approval upgrades them to full demand automatically.
3
application_status — your agent polls until approved; your API token is returned once. Every response embeds an onboarding checklist and a next_action, so your agent always knows what is done, what is next, and who must act.
4
reporting & payments — your agent tracks revenue and traffic, and manages payouts. Monthly, $20 minimum, by PayPal, check, Wise, or Payoneer.
*
recover — if tokens are lost, your agent emails a one-time code to the account owner’s contact address on file, then exchanges it for fresh tokens (old ones are revoked).
// always know what’s next

The checklist & next_action mechanic

Onboarding is a state machine your agent can read. Every application_status response — and the standalone checklist tool — returns the full onboarding checklist: every step to full-demand serving with its status, who must act on it, and concrete instructions. The embedded next_action names the single thing to do next, so your agent never has to guess where it is in the process or poll blindly.

What is done

Each checklist step carries its own status — your agent reads progress instead of inferring it from errors.

What is next

next_action points at the exact step to work on, with instructions concrete enough to execute.

Who must act

Every step names its actor — the agent or the site owner — so nothing stalls waiting on the wrong party.

// integrate once

One tag. Your units serve immediately.

Work with your coding agent to add a single script tag and place ad units wherever they belong. They start serving immediately with placeholder creatives while your application is reviewed — so you can verify placement, sizing, and lazy-fill behavior from the first pageview. When your application is approved, the same units upgrade to full demand automatically; the tag, the units, and your page markup stay exactly as they are.

Agent-native

Every action is an MCP tool: apply, integrate, report, get paid. Your agent runs them in the page after you sign up, or from Cursor, Claude, or the CLI.

Your agent places the ads

Your agent decides where units go, how many, and their size. A bare unit fills its container, or data-rs-width/data-rs-height constrains it. Units fill lazily as they scroll into view and collapse cleanly if nothing fills them.

Reporting & payments

Your agent pulls revenue and traffic reports and manages payouts over MCP — monthly, by PayPal, check, Wise, or Payoneer. Payout method changes and tax forms stay on the secure web dashboard.