Looking for a QA Wolf alternative?
QA Reef is an open-core agentic QA tool you run on your own hardware. It reads the screen with OCR, turns what it sees into real Playwright test code you own, and logs every model decision's exact cost — no opaque "AI credits."
Researched 2026-08-28 from public sources; verify current details at qawolf.com. QA Reef claims are limited to what the repo does today.
Who this is for
Good fit for QA Reef, not QA Wolf's hosted platform, if:
- Self-host your QA automation instead of a vendor's cloud grid.
- Test internal tools, canvas-heavy UIs, or embeds without stable selectors.
- Want every model decision logged and cost-attributed, not an opaque per-credit bill.
- Own the generated test code in your own repo and IDE.
- Don't need a managed human-QA service — your own team triages failures.
QA Wolf is the better fit if you want a fully managed "Coverage as a Service" with dedicated human QA engineers reproducing bugs — its own stated moat, not something QA Reef offers today. Source: qawolf.com/service, accessed 2026-08-28.
Side by side — only what's verifiable
QA Wolf rows are sourced to a public URL; QA Reef rows reflect what's implemented today, not the roadmap.
| QA Reef | QA Wolf | |
|---|---|---|
| Openness | Automation core (runtime, operator, OCR/pixel, LLM adapter, evidence storage) is MIT, built in the open under packages/core. |
Bottom layer only — Playwright/Appium, plus @qawolf/flows (MIT) and @qawolf/cli (Apache-2.0), are public. The platform (Mapping/Automation AI, runner grid, human service) is closed. Source: github.com/qawolf. |
| Where it runs | Your own machine, your own CI, or self-hosted — you choose. | Self-serve flows run on QA Wolf's pre-warmed Kubernetes runner grid. Source: qawolf.com/blog. |
| How it finds things to click | OCR-driven operator (Apple Vision or a vision-model adapter) reads visible text; a cheap model picks the next label. Generated tests use ordinary Playwright locators. | DOM automation for recorded/generated flows; a separate coordinate-based "Runner" exists in the CLI's Agent Skill for external agents. Source: @qawolf/cli npm package. |
| Verdict states | PASS / FAIL / UNMEASURED — a check that couldn't get a clean look is never silently a pass or a fail. | Not a distinct state; a human QA engineer manually verifies before a failure is reported — "zero flake" is human verification, not a third automated state. Source: qawolf.com/service. |
| Model attribution | Every agent-made click/type/assertion is logged with the exact model id and token cost. | Billed by "AI credit," not documented per-call attribution. Source: qawolf.com/pricing. |
| Mobile | Not supported — Playwright/Chromium only today. | Yes, via Appium. Source: github.com/qawolf. |
| Pricing | Not published yet — self-hosted core costs whatever your own model calls cost. See QA Wolf pricing for a sourced look at theirs. | Self-serve: 1¢/AI credit + 15¢/runner-minute, no seat fee. Coverage as a Service: unpublished, billed by tests under management. Source: qawolf.com/pricing. |
Deeper architecture writeup: QA Wolf vs QA Reef, in depth.
How migration actually works
- QA Wolf's
.flow.tsformat (Apache-2.0@qawolf/flows) runs entirely offline via their own free@qawolf/cli. - QA Reef's converter (
app/codegen-flow.js) emits that same shape from a QA Reef-recorded flow — same locator strategy, same secrets convention. - So a flow mapped in QA Reef runs through QA Wolf's free CLI too — no lock-in either way. Verified details:
vendor/QAWOLF-OSS.md. - What doesn't carry over: QA Wolf's closed-platform pieces — Mapping/Automation AI generation, the runner grid, human-verified triage.
Step-by-step version: How to migrate from QA Wolf to self-hosted agentic QA.
Questions people ask before switching
Is QA Reef a drop-in replacement for QA Wolf?
No — QA Wolf's Coverage as a Service (hosted platform + dedicated human QA engineers) isn't something QA Reef offers; QA Reef is for self-hosting and owning your automation.
Do I lose my QA Wolf flows if I switch?
Not necessarily — QA Wolf's .flow.ts format runs offline via its own free CLI, and QA Reef's converter can emit that same shape. Details: vendor/QAWOLF-OSS.md.
What does QA Reef do differently from QA Wolf?
OCR-driven screen reading instead of DOM-only locators, full model-decision cost logging, an MIT open-core, and a PASS/FAIL/UNMEASURED verdict.
What does QA Reef not do yet?
No hosted execution grid, mobile testing, human-verified triage, or self-healing locators yet — see the compare page.
How much does QA Reef cost?
Not published — self-hosting costs only your own model calls, fully cost-logged. See the pricing page and QA Wolf pricing.
Want the open-core details specifically?
What's actually in packages/core, the flow converter, and a self-hosting summary.