37 reports become 11 bugs
Sage groups duplicates. Each bug is one ticket with steps, a clip and the logs.
Put a budget on your next build. Real people play it and report what breaks, Sage scores every report from the clip and the logs, and Claude fixes the confirmed bugs. You read one summary a day.
Reported, confirmed, sent to Claude, fixed, checked by players, verified. You step in only when a card needs you.
This is a real path through BuggyPay on Autopilot. People do the playing and the checking. Sage and Claude do the reading and the fixing. You spend under a minute.
Example campaign: Rust Dunes, map 2. Times are from one test night.
The last 30 seconds, the console logs and the phone model are attached on their own. One sentence: “The worm went through the wall on its second dive.”
Visible in the clip at 00:21, a navmesh warning in the logs, no earlier report like it. Marked S2 Major, three more testers confirm it within the hour.
Held for 48 hours, so a wrong call can be undone.
Steps to reproduce, expected and actual result, the clip and the logs, sent through MCP to your coding agent.
“Clamp the worm dive to the arena bounds.” PR #482, build 0.4.13. Nothing is merged without you.
The tester who found the bug is asked first.
@kitedrop receives the 9.00 USDG bonus. The other confirmations are paid too.
From the daily summary: 23 reports, 17 fixed, 15 verified, 2 waiting for you.
Set it per campaign and change it whenever you like. In every mode Sage explains each score with evidence, and fraud flags, appeals and unsure calls always reach a person.
Every automatic decision can be undone while the payout is on its 48-hour hold. Every action Sage and Claude take is written to an activity log.
Everything a studio needs to run testing as a pipeline, not a pile.
Confirmed bugs go to Claude, fixes come back as a test build, and players check them before you release.
Sage groups duplicates. Each bug is one ticket with steps, a clip and the logs.
Every score lists the evidence behind it, so you can trust it or overrule it in one click.
The BuggyPay MCP server works with Claude Code and any agent that speaks MCP.
Private campaigns run under NDA. Public ones can pay players to post their clips on X.
Pick a mission, play, tap Report when something breaks. Sage tells you your score and why within minutes, and the money lands in your wallet or your bank.
You set the price for each. We start with games, and the same missions work for any app.
Find what is broken. Paid per bug by severity, from S1 Critical to S4 Polish.
e.g. 2.00 to 40.00 USDG per bugA first-time session. Sage marks the moments a new player gets stuck or wants to quit.
e.g. 6.00 USDG per sessionPlay, then answer a few questions on difficulty and fairness.
e.g. 2.00 USDG per survey“We need five cheap Android phones.” Fill the gaps your team cannot test.
e.g. 4.00 USDG per deviceCreated on its own when Claude has a fix ready. Testers confirm it on the new build.
e.g. 1.50 USDG per checkPublic campaigns only. Testers post their clip, and Sage checks the post is real and theirs.
e.g. 3.00 USDG per postPay in USDG or your own token on Robinhood Chain, or in dollars by card through Stripe. Your coding agent is billed by its own provider.
Crypto budgets sit in an on-chain vault with a public address. Testers can see the money is there before they start.
Fund by card and pay testers to their bank with Stripe Connect. Crypto is never required.
Real people find the bugs and check every fix. You approve releases and can switch any campaign back to Manual.
One account per person, caps that grow with rank, a 48-hour hold, and Sage flags reused clips and edited screenshots.
We are opening to a first group of studios and testers soon, starting with games. Join the waitlist and you will be the first to know.
Only inside the rules you set. In Manual your team confirms everything. In Assisted Sage handles small, clear bugs and sends S1 and S2 to you. In Autopilot Sage works within your budget and caps. Every decision shows its reasons and can be undone during the 48-hour hold.
No. Claude opens a pull request and a test build. Merging and releasing stay off until you turn them on, and you approve every release.
No. You can fund by card and testers can withdraw to a bank account. Crypto campaigns use USDG or your own token in an escrow vault on Robinhood Chain.
The same missions work for web and mobile apps. Testers can upload a screenshot or video when there is no SDK in the build.
When you close the campaign, the unused budget comes back to you with its share of the fee.
Anyone aged 18 or over with one verified account. New testers start with small daily caps that grow as their reports are confirmed.
BuggyPay opens to the first studios and testers soon. Join the list to hear first. Studios on the list get early-access pricing.
We will email you as soon as BuggyPay opens.