Java Coverage

This documentation was generated with the assistance of AI. Please report any inaccuracies.

Run JaCoCo (or Kover) against a scoped or full unit-test run of a JVM project — Java or Kotlin sources, Maven or Gradle — then read the generated CSV/XML report to give the exact, measured line-coverage percentage for the specific class(es) the caller cares about. This is a read-only reporting skill: it measures and reports coverage, it does not add tests or modify source/test/build code unless asked to as an explicit follow-up.

Purpose

iru-java-coverage answers "does class X meet the coverage bar?" with an exact, report-derived percentage instead of an eyeballed guess from test output. It detects the build tool and coverage engine (Maven with jacoco-maven-plugin; Gradle with JaCoCo, using a throwaway init script that forces CSV/XML reports on without touching the build; Gradle with Kover; Quarkus with quarkus-jacoco), reuses iru-java-test’s selector syntax to scope which tests run, infers or accepts an explicit list of target classes (including Kotlin’s compiler-generated `FooKt, Foo.Companion and lambda classes), and, for classes under the bar, points to the exact uncovered lines/branches in the per-class HTML or XML report. It measures unit tests only and never runs Failsafe, verify, integrationTest or quarkusIntTest. Android Gradle builds are redirected to iru-android-coverage.

When the Kotlin Multiplatform plugin is applied anywhere in the build, the skill stops and points to KMP Coverage instead.

Purpose group

Java Development Tooling

Plugin

Shipped by the iru-java plugin of the ai-catalog marketplace (install with /plugin install iru-java@ai-catalog; see Plugins).

Invocation

/iru-java-coverage
/iru-java-coverage <selector> [TargetClass1,TargetClass2,...]

Inputs

Argument Required Description Default

selector

No

Which tests to run, using the exact same syntax as the iru-java-test skill’s selector table (class, class+method, comma list, wildcard, package pattern, tag expression). Must cover every test class that exercises the target class(es) — a selector scoped too narrowly undercounts real coverage, so a broader selector (a whole package, or the full suite) is preferred when in doubt.

None — omitting it runs the full unit suite.

TargetClass1,TargetClass2,…​

No

Comma-separated class name(s) to report coverage for.

None — if omitted, target classes are inferred from the test selector by convention (FooTest → Foo), or, for a wildcard/package/tag selector where no name can be inferred, every class touched by the run is reported (or the user is asked which class(es) matter).

Outputs

  • A plain-text summary in the conversation — never a written file.

  • Per target class: exact line-coverage percentage (and branch/instruction coverage where relevant), computed from the CSV or JaCoCo-format XML counters — never estimated — labelled with the engine it came from (JaCoCo Maven, JaCoCo Gradle, Kover, quarkus-jacoco).

  • If the user gave a coverage bar (e.g. "at least 80%"): a clear statement of whether each class meets it.

  • For classes under the bar: a pointer to the per-class HTML report and the exact uncovered line numbers (via the nc/pc/fc span classes, or the XML <line> entries) rather than just the aggregate percentage.

  • A class absent from the report is reported as 0% coverage, not as an error. Kotlin generated-code noise (inline bodies, companion rows, coroutine state machines) is called out as such.

  • Quarkus: @QuarkusTest coverage from quarkus-jacoco and plain-unit-test coverage are reported separately, never summed; classes exercised only by @QuarkusComponentTest without quarkus-jacoco are reported as unmeasured, not 0%.

  • A build, compilation or test failure (including a Docker-less Testcontainers/Dev Services start-up failure) is reported as such instead of reading a stale report.

  • No source, test or build code is ever modified, and coverage is never re-run to try to raise a number.

Execution flow

flowchart TD A["Start /iru-java-coverage\n(optional selector, target classes)"] --> B["Step 0: detect build tool\nand coverage engine"] A -. "KMP plugin applied" .-> KMP["Stop: hand off to iru-kmp-coverage"] B --> B1{Android Gradle plugin?} B1 -- yes --> B2["Defer to iru-android-coverage"] B1 -- no --> C["Step 1: scope tests with\niru-java-test selector syntax\n(unit tests only)"] C --> D["Step 2: given targets, or infer\nFooTest -> Foo (+ Kotlin FooKt, Companion)"] D --> E{Engine} E -- Maven --> F["3a: mvn clean jacoco:prepare-agent\ntest jacoco:report"] E -- Gradle JaCoCo --> G["3b: gradlew clean test --tests\njacocoTestReport + init script"] E -- Gradle Kover --> H["3c: gradlew clean test --tests\nkoverXmlReport"] F --> I{Build succeeded?} G --> I H --> I I -- no --> J["Report failure, stop\n(environment failure if no Docker)"] I -- yes --> K["Step 4: read CSV, else JaCoCo-format XML"] K --> L["Compute line / branch / instruction %\nper target class"] L --> M{Quarkus extension report?} M -- yes --> N["Read quarkus-jacoco report too;\nreport both numbers separately"] M -- no --> O{Class row found?} N --> O O -- no --> P["Report 0% (or unmeasured\nfor component-test-only classes)"] O -- yes --> Q["Step 5: report exact %;\nstate pass/fail vs. any bar"] P --> R Q --> R{Under the bar?} R -- yes --> S["Point to per-class HTML / XML\nfor uncovered lines"] R -- no --> T["State it meets the bar"]

Dependencies

Invokes

None — iru-java-coverage is a leaf skill; it only runs Maven/Gradle commands and reads the generated reports. It reuses the build-tool detection and selector syntax documented in Java Test, and redirects Android Gradle builds to Android Coverage, but it does not call either skill.

It also points Kotlin Multiplatform projects at KMP Coverage but does not invoke it.

Invoked by

  • Java Code One Task Group — Step 3, once per task group’s bucket, via the iru-gate-runner agent, to check every changed/new class in the bucket meets the 80% coverage bar (falling back to mvn jacoco:prepare-agent test jacoco:report directly if this skill is unavailable).

  • Kotlin Spring Boot Code One Task Group — Step 3.5, unit-only 80% coverage gate, via the iru-gate-runner agent.

  • Quarkus Code One Task Group — Step 3.5 and 3.9, unit-stage 80% gate and aggregate report, via the iru-gate-runner agent.

  • Java Generate All Tests — Step 3, and again in Step 6 to re-measure after generation: delegated through the iru-gate-runner agent so its full jacoco.csv/HTML-derived report never lands directly in this conversation, and so the whole suite runs once per batch rather than once per class.

  • Java Spring Boot Code One Task Group — Step 3.5, via iru-gate-runner, measuring the 80% gate over a Surefire-only run. Falls back to mvn clean jacoco:prepare-agent test jacoco:report -DskipITs plus each module’s own jacoco.csv — never the aggregate, which mixes in integration execution data.

Related agents

None — this skill spawns no agents itself; callers run it inside an iru-gate-runner sub-agent.

Source

SKILL.md on GitHub — the file this page was generated from.