Rust Coverage

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

Run a Rust project’s container-free tests under cargo-llvm-cov and report the exact, measured line and region coverage for the files the caller cares about, read from an lcov file and llvm-cov’s JSON summary. It is a read-only measuring skill: it never adds tests, modifies source or test code, changes coverage configuration, or installs tools unless asked to as a follow-up.

Purpose

A coverage number is only meaningful if coverage is actually wired and the right code is measured. iru-rust-coverage checks that cargo-llvm-cov and the llvm-tools component are present, runs the tests once with instrumentation (cargo llvm-cov nextest when cargo-nextest is installed, plain cargo llvm-cov otherwise, always --locked) using --no-report, then writes target/coverage/lcov.info and target/coverage/summary.json and reads both with a small stdlib Python parser. The line % is the gate figure and the region % is reported alongside; branch coverage is reported as unavailable on a stable toolchain, because --branch needs nightly.

Binaries launched by end-to-end tests (for example through assert_cmd) are instrumented too, so a thin src/main.rs is measured fairly. Tooling members such as xtask/ are excluded with --ignore-filename-regex, tests/ files are never reported, and doctests do not count on stable. #[ignore] tests are never run, so the gate value is reproducible without Docker. A run with failing tests still yields a report, labeled as such. The skill never passes --fail-under-* flags: it judges the bar itself. A missing cargo-llvm-cov or llvm-tools-preview, a build failure, or a target absent from the report is reported instead of a false 0%.

Purpose group

Rust Development Tooling

Plugin

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

Invocation

/iru-rust-coverage
/iru-rust-coverage <targets>

Inputs

Argument Required Description Default

targets

No

Comma-separated source files (src/commands/greet.rs), directories (src/commands/), or module paths (crate::output, commands::greet, <member_crate>::config). Module paths are resolved against the owning package’s library root from cargo metadata.

Every reported file in the workspace.

bar: <N>

No

The threshold each target’s line % is judged against.

80

package: <name>

No

Restricts the run to one workspace member (-p <name>).

--workspace

features: all|<comma list>

No

--all-features or --features a,b.

Default features.

filter: <x>

No

Narrows the test run: a test-name substring, or expr:<filterset> (nextest only). A narrow filter undercounts code that other tests also exercise.

The whole suite.

Outputs

  • A plain-text summary in the conversation: the runner (cargo llvm-cov nextest or cargo llvm-cov), the toolchain (stable means no branch data), the selection (--workspace or -p, features, filter, ignored tests not run), what the exclusion regex removed, and whether the tests passed.

  • Total line % and region % (branch reported as unavailable on stable), with a verdict on whether the line % meets the bar.

  • Per target (or per file for a small unscoped run): line % and region %, and for every target under the bar its uncovered line ranges.

  • NOT IN REPORT <target> with the likely cause (module not declared, disabled feature, excluded path, bin-only crate with no executing test) instead of a 0%, and "coverage not wired" with the exact install command when tooling is missing.

  • The files target/coverage/lcov.info, target/coverage/summary.json and target/coverage/run.log are left in place (target/ is gitignored; CI and Sonar read the lcov file). No source, test or configuration is ever modified.

Execution flow

flowchart TD A["Start /iru-rust-coverage\n(optional targets, bar, package, features, filter)"] --> B["Step 0: parse arguments and\nresolve module paths to files"] B --> C["Step 1: check cargo-llvm-cov,\nllvm-tools and nextest"] C --> D{"cargo-llvm-cov and\nllvm-tools present?"} D -- no --> E["Stop: report coverage\nnot wired with the fix"] D -- yes --> F["Step 2: run the tests once,\ninstrumented, with no-report"] F --> G{"Build failed?"} G -- yes --> H["Report the compile error;\nnever read a stale file"] G -- no --> I["Step 3: write lcov and JSON\nsummary excluding tooling members"] I --> J["Step 4: parse lcov and summary for\nexact line and region coverage"] J --> K["Step 5: report totals, per-target\ngaps and bar verdict"] E --> K H --> K

Dependencies

Invokes

None — iru-rust-coverage is a leaf skill. It reuses the runner detection and the container-backed lock procedure (Docker check plus .integration-test.lock) of Rust Test by following the same rules, and borrows that procedure only when integration coverage is explicitly requested.

Invoked by

Related agents

  • iru-gate-runner — calling skills normally run it inside this sub-agent. It spawns no agents itself.

Source

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