The end-to-end testing & quality platform
Build with AI.
Ship with ProvaLab.
AI can build the system. It can't ensure its quality. ProvaLab does — from the first pull request to the release, inside the coding agent you already use.
What ProvaLab is
One pipeline, from the first line of code to the product release
Six stages, in the order your team meets them — and every one of them plugs into the coding agent that wrote the code.
Every clip below is a real recording of ProvaLab on our SelectionSathi demo project, at normal speed — except PR Bot, shown as an illustration.
ProvaLab plugin
Plugs into the coding agent your team already uses
The agent that writes your code can run, generate and triage ProvaLab tests itself, through the ProvaLab MCP server. Claude Code gets a full plugin with skills and slash commands; Cursor, OpenAI Codex, Google Antigravity and Windsurf connect with a few lines of MCP config. Ask in plain language and the agent picks the right ProvaLab tool.
1 · Sign in once per machine
curl -sSf https://plugins.provalab.io/install.sh | sh provalab-auth loginWindows:
irm https://plugins.provalab.io/install.ps1 | iex. Short-lived OAuth tokens; nothing to paste into config.2 · Add ProvaLab to your agent
The plugin for Claude Code, or the MCP entry for the others.
3 · Ask
“Run the checkout tests on ProvaLab”, “why did last night's run fail?” — or
/provalab-testin Claude Code.
Full plugin: skills + slash commands
/plugin marketplace add provalab-dev/claude-code-plugin
/plugin install provalab@provalab-marketplace
# then, in Claude Code:
/provalab-test run a suite, or ask "run the login tests on ProvaLab"
/provalab-status status of a run
/provalab-results results, failures grouped by causeProvaLab's MCP endpoint is streamable HTTP at https://mcp.provalab.io/mcp.
01 · PR Bot
Every pull request, reviewed
Example: a critical SQL injection flagged — and fixed by the coding agent before merge
It starts in your repository. ProvaLab PR Bot reviews every pull request with static analysis, dynamic analysis and security checks, and feeds its findings straight back to your coding agent — so the agent fixes the code before a human ever looks at it.
- Static analysis, dynamic analysis and security checks on every PR
- Findings go back to the coding agent that wrote the code
- Code meets your standards and is not compromised before it moves on
02 · QA Infrastructure
Your agent runs the tests on ProvaLab
A separate UI suite: 14 tests on ProvaLab browsers — 13 passed; the 1 failure is the critical sign-in bypass
Once the code is ready, the tests have to run. Install the ProvaLab plugin or connect through MCP, and your coding agent executes test cases directly on ProvaLab. Browsers, simulators and real devices are provided by the lab.
- Claude Code plugin and MCP: your agent runs the suite itself
- Browsers, simulators and real devices — every session recorded
- One subscription; never manage test infrastructure again
03 · Test Management
Your whole test catalogue, organised
The nightly run in TMS: 95 test cases — 81 passed, 14 failed — and 6 bugs filed from the failures
No more test cases living in spreadsheets. ProvaLab TMS gives you the standard hierarchy — project, module, feature, test suites, test cases, test plans and test runs — and every automated run reports into it.
- Project → module → feature → suite → case → plan → run
- Automated results land on the right case, run by run
- AI-enabled: describe a feature and it drafts the plan, suites and cases
04 · QA AI Agent
Ask your test data anything
“Pass: 156 | Fail: 25” — the same nightly run, counted per automated test, answered in seconds
On top of that sits the QA AI Agent. Ask it in plain language — what did last night's run cover, what failed, what is still open — and it answers from your data. It also executes API and UI test cases on its own.
- Plain-language questions, answered from your own runs and issues
- Runs API and UI test cases when you ask
- Works in the browser and inside your coding agent
05 · AppSec
Find the holes before attackers do
Web app + API scans: missing security headers flagged; 9 false alarms checked and closed
Any attacker can probe a web app that is not hardened. ProvaLab AppSec runs multiple scan types against your app and your APIs, finds the vulnerabilities and makes sure your customers' data stays protected.
- Web app (DAST), API fuzzing, dependencies, secrets, containers and code
- Only targets you have verified you own
- A plain-language fix list, ordered by what matters first
06 · Load Testing
Know your breaking point first
Broke at 25 concurrent users: 55% errors and 502s — the 7th bug of the day, before any customer saw it
The ProvaLab load testing platform generates realistic traffic, so you know how your system behaves under pressure before your customers find out.
- API and real-browser load from the same platform
- Thresholds that pass or fail a run, with a timeline of what broke
- Engine health shown with every run, so results are never a guess
Works with your stack
Keep your languages, frameworks and tools. ProvaLab plugs into them; it doesn't replace them.
Languages
Testing Frameworks
Project Management
Version Control & CI/CD
Communication
You decide where your code runs
Giving an agent your repo and your test infrastructure means trust is non-negotiable. Here's how we handle yours.
Enterprise: your code stays in your network
Enterprise plans run the agent inside your VPC or on-prem; your repositories are never copied to our servers. SaaS today; private VPC and on-prem for Enterprise.
Private, dedicated infrastructure
Enterprise deployments get their own device farm and execution environment — no shared tenancy with other customers.
SSO and role-based access
Connect your existing identity provider and control who can view projects, trigger runs, and provision devices.
Starter and Team are self-serve on ProvaLab's cloud. Enterprise runs in your VPC or on-prem and is set up with you directly. Book a 30-min demo and we'll walk through it against your own stack.
Simple, transparent pricing
Start free. Scale as you grow. No hidden fees.
Starter
Core test management for small teams getting started.
$0
- Up to 3 users
- 500 test cases
- Basic test management
Team
AI agents, device access, and full test management for growing teams.
$79 /user/month
- Up to 25 users
- Unlimited test cases
- Full AI QA Agent access ($100/month AI budget)
Enterprise
On-premise deployment, private device farms, and custom solutions.
Custom
- Unlimited users & test cases
- Dedicated AI agents
- Private device farm (real + virtual)
One subscription instead of two — and less triage. A Team seat covers test management, cloud browsers and devices, and the AI QA Agent: work that usually means a test-management tool, a separate browser-grid subscription, and someone reading through failures by hand.
See it run against your own app.
30 minutes, no slides. We test one of your flows live and you leave with a plan.
- · You describe one flow you care about.
- · We generate and run the tests on real browsers or devices, live.
- · You get a written plan: what to migrate, what the agent writes, what it costs.
Prefer a quick question? [email protected]