You run our agent.Software
Two layers. Same agent.
The repo is the QA agent we use on every engagement, free, for your own Cursor or Claude Code. Qaido is the same agent running on its own across your QA board, as a paid SaaS, with a human signing every result.
The agent we test with. Yours to fork.
Drop it into your repo and your coding agent starts thinking like a QA engineer: it turns acceptance criteria into test cases, writes automation in your framework, and refuses to mark anything done without a run attached.
defect-taxonomy
golden-task-regression
exploratory-charters
automate-in-your-framework
automation-engineer
blind-reviewer
ticket-writer
no-done-without-run
Skills
Acceptance criteria to test cases, defect taxonomy, golden-task regression, exploratory charters, automation in your own framework.
Subagents
A test designer, an automation engineer, a ticket writer, and a blind reviewer with a mandatory "Not verified" section.
Hooks
Rules that run outside the model. No destructive action on tracked artifacts. No ticket to Done without a run attached.
Qaido. One agent, the whole QA board.
Connect Jira and GitHub. The agent picks up the next ticket in the QA column, plans the tests, writes them in your framework, runs them, files what breaks, and hands the result to your QA lead to sign. Highest priority first, unattended.
Get early access at qaido.aiDestructive actions are denied
On tracked artifacts: tests, golden tasks, tickets. The hook runs outside the model. It is not a prompt asking nicely.
Nothing moves to Done without a run
Every ticket the agent closes has a run artifact attached: trace, video, results. Or it does not close.
A human signs
The agent stops at sign-off. Your QA lead reads the verdict and the "Not verified" list, and decides.
What it will not become.
Runs in your infrastructure by default, or in a managed sandbox. At launch: Jira, GitHub, Qase.
Get early access