Start free
$0
20 credits when you sign up — enough for your first patch check.
Your coding agent reads each patch, decides what it puts at risk, and hands the build to real human playtesters. You get back a sentence and a verdict — with the session on video — before you ship.
you ▸run QA on this patch
agentdiff touches saves and the pause menu — 2 checks affected
agentbuild published, playtest rounds filed. working on…
agentverdict: old saves no longer load. issue filed, video at 04:12
How it works
npx @qaping/cli setup connects qaping to Claude Code, Cursor, or Codex — the same agent you already build with.
Ask it to set up QA. It reads your repo and drafts a committed checklist of what your game can't afford to break — saves that must load, the first boss, the settings screen. You edit it like any other file.
The agent reads the diff, picks the checks that patch touches, publishes the build, and files playtest rounds with real people from the pingfusi reviewer pool. Verdicts, questionnaire answers, and full session video land on your PR — issues filed and timestamped to the footage.
When a check has settled — same result, patch after patch — your agent promotes it into a plain code test in your repo, free to run on every PR. Replaying recorded sessions on a machine is in pilot, not self-serve yet. Human rounds stay for what only judgment can answer: feel, readability, whether the boss is actually fun.
qaping is an MCP server your coding agent calls. One command wires it up.
Type it in your terminal, or in Claude Code, Cursor, or Codex. It sets everything up for you.
npx @qaping/cli setupSign in when prompted.
Your agent authors QA-PLAN.md from your repo. From then on, “run QA” is the whole workflow.
set up QA for my game. use qapingPay for a playtester's time, per round. No subscription.
Start free
$0
20 credits when you sign up — enough for your first patch check.
Then per round
One command to set up. The free credits are enough for your first patch check.