Julia Test ItemsA modern testing framework for Julia

Write standalone test items, run them anywhere — VS Code, REPL, command line, or CI.

VS Code with test items, inline results, and code coverage

Standalone Test Items

Structure tests into independent @testitem blocks that can be run individually. No more running your entire test suite to check one thing.

Run Anywhere

The same tests work in VS Code, the REPL, on the command line, and in GitHub Actions CI — no changes needed.

Parallel Execution

Test items run in parallel across multiple worker processes, with automatic process management and configurable concurrency.

Code Coverage & Debugging

Run tests with code coverage on Julia 1.11+, or step through them in the VS Code debugger with full breakpoint support.

Built for AI Agents

An MCP server exposes the same discovery, execution, and diagnostics to coding agents, so they run individual test items and read structured failures instead of scraping output.

Rich VS Code Integration

Discover test items automatically as you type. Run them individually with a click and see results inline — failed assertions show expected vs. actual values right in the editor. Line-level code coverage highlights which code paths your tests exercise.

VS Code showing test explorer, inline test failure diffs, and line-level code coverage annotations

A Real Command Line Runner

The juliati command finds every test item under a folder and runs them in parallel test processes — no editor and no runtests.jl required. Filter by name or tag, measure coverage, and write machine-readable results for CI. (Prerelease)

$ juliati
  Discovered 24 test item run(s) in 3 file(s)
  Launching test processes....
  Progress: 24/24 (23 passed, 1 failed)

24 tests ran, 23 passed, 1 failed.

Interactive REPL Mode

DevREPL.jl adds a dev> mode to the Julia REPL. Pick test items from a fuzzy list, rerun just the failures, and browse them interactively — all without leaving the terminal. (Prerelease)

julia> )

dev> test results
Run #2: 184 test(s) (60.6 s) — 183 passed, 1 failed

  [FAIL] get_name (1388.9ms)
    Test Failed at test/test_interface.jl:68
      Expression: valof(get_name(parse("struct T end"))) == "F"
       Evaluated: "T" == "F"

dev> test failed

Batteries-Included CI

A single reusable workflow gives you linting, format checking, testing across a full Julia version and platform matrix, coverage upload, a rich job summary, documentation deployment, and TagBot — all configured in one YAML file. Prefer to build your own pipeline? Each step is also a standalone action.

GitHub Actions workflow showing lint, test matrix, report results, and automation stages

Test Items for AI Agents

JuliaMCP.jl is an MCP server that hands coding agents a live Julia environment. An agent can run a subset of test items, read the structured failures, lint a file, and evaluate code in a persistent session — instead of shelling out to julia and scraping stdout. (Prerelease)

json

{
  "mcpServers": {
    "julia": {
      "command": "juliamcp"
    }
  }
}

Simple to Write

Define independent test items anywhere in your package. Each one is self-contained and runnable on its own.

julia

@testitem "CSV parsing" tags=[:parser] begin
    data = parsecsv("name,age\nAlice,30\nBob,25")

    @test length(data) == 2
    @test data[1].name == "Alice"
    @test data[2].age == 25
end

The Test Item Ecosystem

TestItems.jl — The core @testitem, @testmodule, and @testsnippet macros.
Julia VS Code Extension — Rich editor integration with inline results, debugging, and coverage.
TestItemApp.jl — The juliati command line runner. (Prerelease) — guide
DevREPL.jl — The dev> REPL mode, with a test item picker and failure browser. (Prerelease) — guide
JuliaMCP.jl — MCP server giving AI coding agents a live Julia environment. (Prerelease) — guide
testitem-workflow — Reusable GitHub Workflow for CI with version matrix support — guide
GitHub Actions — The individual actions the workflow is built from, usable on their own.
TestItemRunner.jl — Runs test items under Pkg.test, for compatibility with the traditional workflow. — guide

Under the hood, discovery is powered by JuliaWorkspaces.jl and TestItemDetection.jl, and execution by TestItemControllers.jl — the same engines every surface above shares. TestItemRuns.jl packages them into one Julia API for building your own tools; see Integrating.

Read the Guide →