Product
Solutions
Resources
Company
Download Trial Book a Demo
RKTracer · Code coverage for every language & target

Reliable code coverage.
Every language. Every target.

RKTracer measures every coverage metric, including MC/DC, multi-condition and delta, across embedded targets, GPUs, simulators and hosts, without changing your build. Then it hands the uncovered lines and conditions to your AI agent, so the AI writes the tests that close the gaps.

Coverage that fits how you already build & test

No source changesJust prefix the build Every compiler& cross-compiler Host → targetGPU & CUDA too

One coverage engine for every place your code runs

Embedded targets GPUs & CUDA Simulators Emulators Host machines Servers & cloud Android devices RTOS & bare-metal
Why RKValidate

AI can write tests. It can't see what's still untested.

Coverage is the ground truth AI test generators are missing. RKTracer measures every metric, through MC/DC, on the code as it actually runs, then hands your AI agent the exact uncovered lines and conditions over MCP. RKTracer measures the gaps; your AI writes the tests that close them.

Measure on the real target

Coverage on the actual silicon: Cortex-M, GPU & CUDA, DSP, RTOS or bare-metal, host or target, with or without a file system, across every compiler and cross-compiler.

Every metric, through MC/DC

Function, file, statement, line, decision, condition, multi-condition, MC/DC and delta: the structural metrics functional-safety standards require, and most tools never measure.

Feed the gaps to your AI

The built-in RKMCP server exposes your covered and uncovered lines and conditions as JSON-RPC over MCP, so your AI agent can generate the unit tests that close them.

RKTracer vs. typical coverage tools

Coverage that follows your code all the way to the target

Most coverage tools stop at line and branch coverage on the host. RKTracer measures the deeper structural metrics, including MC/DC, multi-condition and delta, on the hardware your code actually runs on, and turns the gaps into data your AI can act on.

  • MC/DC, multi-condition & delta, not just statement %
  • Runs on embedded targets, GPUs & simulators, not only the host
  • Feeds uncovered lines & conditions to your AI agent via MCP
CapabilityTypical coverage toolRKTracer
SetupWrappers, env, build editsPrefix the build
Source changesSometimesNone
MC/DC & multi-conditionOften missingNative
Embedded / GPU / simulatorHost-focusedAll targets
CUDA coverageNoYes
Delta coverageRareYes
AI-ready gap data (MCP)NoYes

Comparison reflects common capabilities of mainstream coverage tools; individual tools vary.

Complete coverage

Every metric the standards ask for

From the basics to the hardest structural metric in software assurance.

Function, File & Line

Know exactly which functions, files and lines your tests execute.

Statement & Branch/Decision

Confirm every statement runs and both outcomes of each decision are tested.

Condition & MC/DC

The strongest structural metrics, the ones the highest assurance levels require.

Multi-Condition & Delta

Go beyond MC/DC, and focus coverage on exactly what changed each build.

Languages

10 languages. One coverage standard.

From bare-metal C to modern Rust and Go, RKTracer speaks your stack, and applies the same rigorous coverage model to all of it.

See all supported languages & compilers
CC C++C++ CUCUDA RsRust C#C# JvJava JSJavaScript TSTypeScript GoGo PyPython
How it works

From your build to coverage and AI-ready gaps

Prefix your build

Add the rktracer keyword to your existing build. It works with any build system and any compiler or cross-compiler. Source is instrumented automatically; nothing else changes.

Run your tests, any way

Run unit, integration, functional or UI tests on host, simulator, emulator or the real target (incl. GPU & CUDA). The runtime is linked in automatically, even when cross-compiled.

Get the coverage report

Function through MC/DC, multi-condition and delta, including on-device coverage, with low overhead, in HTML and XML.

Hand the gaps to your AI

The RKMCP server serves the uncovered lines and conditions as JSON-RPC over MCP, so your AI agent can write the unit tests that close them.

How RKTracer works
your build
$ rktracer make
compiler auto-detected, source unmodified
instruments at source
Run your tests, any target
HostEmbeddedGPU / CUDASimulator
captures every metric
Coverage report
Statement100%
Decision96%
MC/DC88%
sends uncovered gaps over MCP
Your AI agent writes the tests
✓ test_brake.c✓ test_flight_law.c✓ test_can_rx.c
Jenkins
Azure DevOps
GitLab CI
SonarQube
HTML / XML
Bamboo
From coverage to tests

Two ways to turn uncovered code into tests

RKTracer finds exactly what is untested, through MC/DC. Then you choose how to close the gaps: hand them to the AI agent you already use with RKMCP, or generate the unit tests fully offline with RKTracerGen. Both are part of the RKTracer tool.

RKMCP AI
AI-assisted test generation

Stream the uncovered code to the AI agent you already use, and let it write and run the tests.

1
RKTracer finds the gaps
Uncovered lines, decisions, multi-condition and MC/DC conditions.
2
RKMCP serves them to your AI
A C++ MCP server streams the gaps as JSON-RPC, in memory, into your agent.
3
The AI writes and runs the tests
Unit and functional tests, plus the Makefile or build, compiled and executed.
Gaps closed
It re-runs until the once-uncovered code is covered.
Best for
Teams that already use AI in their workflow.
RKTracerGen Offline
Deterministic, no AI

Point it at one function or a whole file. It writes, builds and runs the unit tests, with no AI and no network.

1
Pick a function or file
Generate for a single function, or a whole source file.
2
It designs the tests offline
Boundary-Value Analysis, then runs your code once to capture the real expected values, and stubs out every call made inside the function.
3
Builds and runs locally
Emits a Unity, GoogleTest or doctest harness, managed stubs and a standalone Makefile, then builds the binary and runs it.
Gaps closed
Unit tests pass and coverage is reported locally. No tokens, no orchestrator.
Best for
Fast, repeatable, air-gapped or CI runs.
At a glanceRKMCPRKTracerGen
Needs AIYour AI agentNo, fully offline
Tests it producesUnit and functionalUnit, boundary values
Calls inside the unitHandled by the AIStubbed by default
What you getAI-written tests and buildHarness, stubs and Makefile
Build and runThe AI compiles and runsBuilds and runs the binary locally
Running costAI tokens and computeNone, no tokens
ScopeAcross the project via the agentPer function or per file

Both build on the same RKTracer coverage engine, and suit safety-critical workflows that follow IEC 61508, ISO 26262, EN 50128 or DO-178C.

10
Languages supported
9
Coverage metrics, through MC/DC
100%
Compiler & cross-compiler support
0
Build-workflow changes required
What our customers say

Teams have shipped with RKTracer for years, not weeks

Engineering teams in medical devices, climate systems and embedded software have relied on RKTracer for three to four years. A few words from them.

"We use RKTracer to get our test-coverage matrix across both unit and functional or integration testing. The coverage gaps it surfaces let us write better tests and catch bugs that would otherwise stay hidden, and it even helps us spot dead code. On-target C coverage is something our earlier Python tooling could not match, and any issue we raised has been resolved in a software update."

OB
Or B.
Medtronic · Medical devices
3+ years

"RKTracer streamlined our coverage analysis with precise metrics that pinpoint untested paths, which has raised our test quality and our confidence in every release. It uncovers gaps other tools missed, even across our BSP layer and multi-threaded modules, and it slots cleanly into our CI/CD pipeline. The accuracy, the reporting and the support have all been excellent."

BP
Bala P.
Lennox · HVAC and embedded
3+ years

Measure real coverage. Let your AI close the gaps.

Try RKTracer free on your own code, host or target, and see full MC/DC coverage in an afternoon.