Skip to main content
Interactive background network visualization showing connected nodes representing system dependencies.
About BitDive

We Make Code Changes Verifiably Safe

BitDive captures real behavior from running Java and Kotlin services, then auto-generates deterministic JUnit tests that prove your changes work. No mock scripts. No guesswork.

The Verification Platform for Modern JVM Teams

Every day, engineering teams push code changes and hope they don't break anything. They write manual tests based on assumptions, maintain thousands of fragile mocks, and still miss regressions in production.

BitDive eliminates that uncertainty. We attach a lightweight agent to your Java or Kotlin services that records exactly what happened—every method call, SQL query, HTTP request, and Kafka message. Then we turn those recordings into executable JUnit tests that replay the exact same behavior.

The result: you verify every code change against what actually happened in your system, not what someone imagined might happen.

Capture

Zero-code agent records live execution context from any environment

Generate

Auto-create isolated JUnit tests with real data, real mocks, real assertions

Verify

Replay in CI/CD to detect regressions before they reach production

35.5K+
Maven Downloads
1.1K+
Active Installs
117+
Trusted Teams
0
Lines of Test Code

From Running Service to Verified Test Suite in Minutes

No SDK integration. No test scripts. Attach the agent, record behavior, generate tests.

01

Attach & Record

Add the BitDive Java agent to your service. It captures every method call, SQL query, HTTP request, and message queue interaction with minimal overhead.

02

Select & Generate

Browse captured traces in the visual Service Map. Select any execution flow and click "Generate Test." BitDive creates a deterministic JUnit test with auto-mocked dependencies.

03

Run & Verify

Run the generated tests with standard mvn test. They detect behavioral regressions—changed SQL patterns, altered API responses, broken call sequences—not just compilation errors.

Built by Engineers Who Were Tired of Writing Mocks

We spent years building and testing complex distributed systems—Spring Boot microservices talking to Kafka, PostgreSQL, Redis, and dozens of internal APIs. We wrote thousands of mock-based unit tests. And we watched them fail to catch the bugs that mattered.

The problem was clear: mock-based tests verify assumptions, not reality. When a Kafka consumer changes its serialization format, when a SQL query returns a different column order after a migration, when a third-party API adds a new field—your mocks don't know.

"What if tests could replay what actually happened instead of what we imagined would happen?"

That question became BitDive. We built a platform that captures real runtime behavior and converts it into deterministic, isolated test scenarios that run anywhere without external dependencies.

2024

Platform Launch

First public release of the BitDive agent and dashboard. JUnit test generation from recorded traces.

2024

Service Map & Call Tree

Visual distributed tracing with deep call-tree inspection and method-level argument recording.

2025

AI Runtime Context (MCP)

Exposed JVM execution context to AI coding agents via Model Context Protocol. Cursor, Claude, Windsurf integration.

2025

35K+ Maven Downloads

Crossed 35,000 Maven Central downloads and 100+ trusted engineering teams.

2026

Integration Testing

Full-stack integration testing for Spring Boot services. Full context, auto-stubbed boundaries from captured traces.

How We Think About Quality

These aren't abstract values. They're the technical decisions we make every day.

Reality Over Assumptions

Tests should be grounded in observed behavior, not imagined scenarios. Captured data is more reliable than hand-written fixtures.

Determinism by Default

Every generated test produces the same result on every run. No flaky tests, no environment-dependent failures, no "works on my machine."

Zero Code for Developers

The best test automation is the one you don't have to write. We auto-generate test code, mocks, and assertions from recorded behavior.

Standard Tooling

Generated tests are pure JUnit 5 code that runs with Maven or Gradle. No proprietary runners, no vendor lock-in, no new CLI to learn.

Minimal Overhead

The recording agent adds single-digit millisecond overhead. It's designed for staging and production, not just test environments.

AI-Native Verification

AI agents should verify their own code against reality. We provide runtime context via MCP so agents can self-correct before changes ship.

Built for the Modern JVM Stack

BitDive supports the frameworks and tools your team already uses.

Java 8–21 - Official technology supported by BitDiveJava 8–21
Kotlin - Official technology supported by BitDiveKotlin
Spring Boot - Official technology supported by BitDiveSpring Boot
JUnit 5 - Official technology supported by BitDiveJUnit 5
Maven - Official technology supported by BitDiveMaven
PostgreSQL - Official technology supported by BitDivePostgreSQL
MongoDB - Official technology supported by BitDiveMongoDB
Kafka - Official technology supported by BitDiveKafka
Testcontainers - Official technology supported by BitDiveTestcontainers

Common Questions

What does BitDive do?

BitDive captures real execution behavior from running Java/Kotlin services and auto-generates deterministic JUnit tests. Tests replay actual SQL queries, HTTP calls, and Kafka messages—no manual mocking required.

How is it different from Mockito?

Mockito requires you to guess what production data looks like. BitDive records real data from running services and generates the mocks, assertions, and test structure automatically.

Do I need to change my code?

No. BitDive uses a Java agent that attaches to your service at startup. No SDK, no annotations, no code modifications. Generated tests are standard JUnit 5.

What about performance overhead?

The recording agent adds single-digit millisecond overhead per request. It's designed for staging and production environments, not just local development.

Can AI agents use BitDive?

Yes. BitDive exposes runtime context (call stacks, SQL queries, variable states) to AI coding agents via the Model Context Protocol (MCP) for tools like Cursor and Claude.

What JVM versions are supported?

Java 8 through 21, Kotlin, Spring Boot 2.x and 3.x, and all major databases and message queues in the JVM ecosystem.

See It Working in Your Environment

Attach the agent. Record your first trace. Generate JUnit tests in 5 minutes.