Explore
Inspect the start page, then spend the remaining page budget on useful, different internal routes.
An AI website tester that explores on its own
Explore. Investigate. Report.
Give CheckQuest a website. It explores on its own to look for potential problems you did not already tell it to test for, safely investigates what it finds, and gives you an evidence-backed QA report. AI can suggest what looks suspicious; CheckQuest controls what may be investigated and what the evidence actually proves.
Experimental · CLI today · local desktop app next
Start
Home page
Explore
New route
Candidate
Possible issue
Evidence
Ready to report
How a run works
From first pass to final report.
Investigate
Take only narrow, approved actions that can gather evidence for a specific potential issue.
Report
Reconcile what happened into explicit verified or inconclusive findings with supporting evidence.
More than an LLM with a browser
AI can propose. Evidence has to be earned.
CheckQuest uses a model for exploratory judgment, but the model is not the authority on what is true. The QA system around it controls scope, interaction, evidence, and finding state.
The model proposes
Which route, observation, or page-local candidate looks worth spending a bounded exploratory budget on?
CheckQuest constrains
Deterministic budgets, host policy, candidate-linked actions, and safety boundaries decide what the browser may actually do.
The evidence decides
Repeated observations are reconciled into explicit verified or inconclusive findings. A successful browser action is not proof by itself.
Where it fits
A different job in the QA toolbox.
CheckQuest is not trying to swallow the rest of your test stack. It fills one specific gap: a repeatable autonomous exploratory pass.
Often confused with
Site crawler
Broad inventory and exhaustive rule scanning.
Generic browser agent
Gives a model broad browser authority and treats its output as the result.
Test-script generator
Creates or maintains predefined automated checks.
Pentest tool
Actively probes security weaknesses.
What it is
An autonomous exploratory QA pass.
Start from a URL. Spend a bounded exploration budget. Investigate potential issues only when an approved action can gather evidence. Report what the evidence actually supports.
The model proposes. CheckQuest constrains, investigates, verifies, and reports.
Built to complement Playwright, Cypress, Selenium, site-quality tooling, and human exploratory testing — not replace them.
Safety by design
Autonomous, not unbounded.
AI can help decide what looks interesting. Deterministic code decides what the browser is actually allowed to do.
Bound the run
Explicit page and navigation budgets keep the exploratory pass finite.
Constrain interaction
Autonomous actions stay narrow, candidate-linked, and purpose-specific.
Back off when sensitive
Sensitive contexts reduce authority instead of inviting more experimentation.
Keep security passive
Observe useful security posture without turning the run into vulnerability probing.
Right now
The engine exists. The shell is catching up.
Available
CLI · arbitrary authorized URLs · Gemini API key (BYOK) · structured reports
Next
Local desktop product shell