untested
untested
Section titled “untested”Coverage tools answer this question by running the tests and watching what executes. This command answers it without running anything: a test file is a root, the files its imports reach are the code the tests can execute, and a symbol is exercised when a reached file names it. What no reached file names is untested — not uncovered-by-a-run, but unreachable from every test in the tree, which is the stronger and the cheaper claim.
sephera untested --path . # every declared symbol no test reachessephera untested --path crates/sephera_core # the same question, one cratesephera untested --path . --format json # the same list, machine-readableRun from inside the repository you want to measure.
Basic usage
Section titled “Basic usage”sephera untested --path .sephera untested --path . --format jsonsephera untested --path . --fail-on 10The two tiers
Section titled “The two tiers”A symbol a test file names itself is exercised directly: the test calls it on purpose. A symbol named only by a file the tests reach is exercised indirectly: it runs when the tests run, but no test chose it. Both count as tested, and the report keeps the tiers apart because they mean different things — the direct tier is where somebody wrote a test for something, and the indirect tier is where coverage is an accident of the call chain.
What counts as a test file
Section titled “What counts as a test file”The conventions ecosystems agree on for separate test files: a
test, tests, __tests__, spec, or specs directory anywhere
in the path, or a file named test_*, *_test.*, *.test.*, or
*.spec.*. Fixture helpers and configuration match none of these,
so they are not roots: a helper a test imports is reached from the
test, not a test itself.
What does not count
Section titled “What does not count”- A namespace import names nothing, so it is never evidence. A test importing a module wholesale may use any of its symbols without naming one.
- Symbols declared in a test file itself are not code under test: a test function is not something another test needs to cover.
- Module-level values and import bindings are aliases and constants, not behavior a test can exercise, so they are not listed.
- Only Rust and Python collect declarations, so only their symbols can be reported.
The limit, stated plainly
Section titled “The limit, stated plainly”Inline unit tests do not cross an import edge. A Rust #[cfg(test)]
module inside lib.rs tests its own file’s symbols without
importing them, so those symbols read as untested unless a separate
test file names them. This command measures coverage through the
import graph, the way a new test file would see the code; it does
not read test bodies.
A gate on coverage
Section titled “A gate on coverage”--fail-on COUNT exits with code 2 when at least that many symbols
are exercised by nothing. The report is still printed; only the
exit code changes, so a pipeline can gate on test coverage without
running the tests.
SARIF export for code scanning
Section titled “SARIF export for code scanning”--format sarif renders the same report as SARIF 2.1.0, so a
pipeline can upload it to GitHub code scanning and get each
untested symbol as a PR annotation instead of a log line:
sephera untested --path . --format sarif --output results.sarif- name: Run Sephera untested run: sephera untested --path . --format sarif --output results.sarif- name: Upload SARIF uses: github/codeql-action/upload-sarif@v4 with: sarif_file: results.sarifLocations are file-level only: the report carries file paths, never line numbers, so no line is ever guessed — a result points at the file that declares the untested symbol. GitHub accepts file-level results. The direct, indirect, and untested counts ride on the run’s properties rather than repeating on every result.
No test reaches here symbols the suite never names gaps awaiting tests