Action Credits, Explained: Budgeting an SEO Agent
Why yeet charges for actions instead of seats, what each action costs, and a worked month for a 3-site Pro user.
Most SaaS pricing charges for seats — how many people can log in. That model doesn't fit a tool where the main "user" is the agent, not a person. yeet charges for actions instead: things it actually does on your behalf. Seats are practically irrelevant here, since one person can run yeet across ten sites just as easily as ten people running it across one each.
The price list
| Action | Cost |
|---|---|
| Check (a single watcher pass — run_check) | 1 credit |
| Investigation (findings reasoned into a card) | 3 credits |
| Auto-apply PR (yeet opens a pull request) | 2 credits |
| Hands-off verified fix (auto-merged AND verified) | 5 credits |
That last row is worth reading twice, because both conditions have to be true. It only fires when a fix was auto-merged — meaning your auto-merge switch was on and the fix cleared the confidence, fix-type, and CI rules covered in the autonomy modes guide — and later verified, meaning it moved its predicted metric per how yeet verifies fixes. A PR that opens and sits waiting for your review costs the 2-credit auto-apply charge and nothing more; the 5-credit charge is reserved for the fully hands-off path, where yeet acted and then proved it worked with no human step in between.
What's free
Indexing submissions — getting a URL in front of Google via URL inspection and the Indexing API — don't cost credits, as long as you're within your plan's daily submission cap. yeet still respects Google's own per-project daily caps and retries on a 12/30/60-day backoff schedule, up to four attempts, but none of that touches your credit balance. See get indexed on Google fast for how that retry logic works.
Plans and included credits
| Plan | Price | Sites | Submissions/day | Credits/mo | Notes |
|---|---|---|---|---|---|
| Free | $0 | 1 | 10 | 5 (starter) | No auto-apply |
| Pro | $29/mo | 3 | 200 | 60 | MCP/API access, auto-apply |
| Agent | $99/mo | 10 | 1000 | 300 | Aggressive autonomy |
If you run out of included credits mid-month, top-ups run $0.30–$0.50 per credit rather than forcing an upgrade. As of mid-2026, existing users are grandfathered into their current terms for 60 days from mid-July 2026, so if this pricing structure is newer than your account, you have a window before it applies to you.
A worked month: 3-site Pro user
Say you're on Pro, running yeet across three sites, autonomy set to auto_apply_safe, auto-merge off — a fairly typical setup for someone who trusts the safe fix types but still wants eyes on every PR before it merges. A rough month, purely illustrative, might look like:
- Checks: watchers run constantly in the background, but the ones that turn up nothing are the cheap end of the ledger — call it roughly 20–30 credits across three sites over a month, once you net routine polling against the checks that surface something
- Investigations: maybe 8–10 across three sites in a normal month, at 3 credits each — 24–30 credits
- Auto-apply PRs: maybe 4–6 PRs open automatically for safe-type fixes at 2 credits each — 8–12 credits
- Hands-off verified fixes: zero, since auto-merge is off — nothing merges without you, so this line never fires
That lands somewhere in the 55–70 credit range against a 60-credit monthly allowance — close enough that an active month across three sites can bump into a top-up, especially if investigations run heavier than usual. Turning auto-merge on wouldn't reduce this; if anything it adds occasional 5-credit hands-off charges on top, in exchange for not clicking merge yourself. The real math depends entirely on how much yeet actually finds — a quiet month on stable sites might use half of this.
When Agent pays for itself
The jump from Pro to Agent, $29 to $99, buys three things at once: room for 10 sites instead of 3, a 1000/day submission cap instead of 200, and 300 credits instead of 60. The credit jump is the one that matters most for anyone running the math above across more than three or four sites, or leaning hard on auto-apply where investigations and PRs stack up fast — 60 credits disappears quickly once you're past a handful of active sites with real traffic and real findings. If you're consistently topping up on Pro by more than roughly $20–30 a month, or you need more than Pro's 3-site cap at all, Agent's flat $99 with five times the credits is probably already cheaper than Pro plus top-ups. If you're running one or two quiet sites where a month rarely produces more than a handful of investigations, Pro's headroom is probably plenty, and Agent buys capacity you won't use yet.
Where to check your own numbers
Your actual credit usage sits alongside your plan balance on the pricing page, and the docs cover how run_check and the other action types report back — every check returns a small status object (found, did, awaiting_approval, verified_since_last) so you can see, action by action, what a credit actually bought.
Quick FAQ
Does a card that never gets applied cost anything beyond the investigation? No — the investigation that created it already cost 3 credits; a card sitting unapplied in your queue doesn't accrue further charges.
What if I reject an auto-applied PR — do I get the 2 credits back? The PR-open action already happened and used the resources it costs to generate; credits are charged for the action taken, not for whether you agree with the outcome afterward.
What exactly counts as the "hands-off" 5-credit charge? Both conditions have to hold: the fix was auto-merged (autonomy and auto-merge thresholds met) and it was later verified by yeet's metric check. A PR that opens but needs your manual merge, or one that merges but never verifies, doesn't hit this charge.