JavaScript tests are found only by file name (*.test.*, *.spec.*, __tests__/): the classic test/ directory with mocha, node --test, node:assert is "no tests"; jest via node ./node_modules/jest/bin/jest.js, borp, xo, nub are unknown runners

The JavaScript test and runner detection lives in ecosystems/javascript/init.py: the file-name regex near line 309, the _JS_RUNNERS set near line 311, and the _js_tests logic around lines 314-323. Start by reproducing one case from the issue, such as a test/index.js run by npm test, and check which branch rejects it. Done means the listed repros (mocha in test/, a node <path>/jest/bin/jest.js script, and borp) pass with fixtures and the existing tests still pass.

express, got, mime-db (whose nyc coverage PASSes in the same run), octokit/graphql-schema: a test/ directory run by npm test → FAIL "no tests found". phoenix and actions/checkout run node --experimental-vm-modules ./node_modules/jest/bin/jest.js (Jest's documented ESM invocation) → "no CI runs jest … (or npm test)" although npm test was read. fastify runs borp (a node:test runner wrapping c8, with --check-coverage --lines 100), got runs xo (eslint wrapper, thousands of packages), zod runs everything through nub run <script>. ecosystems/javascript/__init__.py:309 should also take test/**/*.{js,ts} (and tests/) when a known runner or npm test runs, and match runners by the script they execute. Also: got's tsconfig extends a strict base, so the "enable strict" hint is wrong.

Evidence from the corpus audit

phoenixframework/phoenix / tests-run

• kind: FALSE_NEGATIVE (MESSAGE)
• actual: FAIL "JavaScript/TypeScript: tests found in 6 file(s), but no CI configuration runs jest, vitest, mocha,
  node --test, deno test, bun test or playwright (or npm test)"
• expected: PASS for JavaScript: CI runs npm test, whose script runs Jest.
• evidence: .github/workflows/ci.yml:151: run: npm test; package.json:35: "test": "node --experimental-vm-modules ./node_modules/jest/bin/jest.js"; jest.config.js:21: testRegex: "/assets/test/.*_test\\.js$". The facts show the
  script was expanded (Command origin package.json:scripts.test argv ["node","--experimental-vm-modules", "./node_modules/jest/bin/jest.js"]), but project_validator/ecosystems/javascript/__init__.py:314-323 only
  accepts program(argv) in {"jest","vitest","mocha","ava"} or node --test; a runner launched as
  node <path>/jest/bin/jest.js (the documented way to run Jest with ESM) is not recognised. The message's
  "(or npm test)" is also misleading: npm test was run and read.
• repro: package.json {"scripts":{"test":"node --experimental-vm-modules ./node_modules/jest/bin/jest.js"}},
  a.test.js, workflow run: npm test
• confidence: high

jshttp/mime-db / tests-run

• kind: FALSE_NEGATIVE
• actual: FAIL: no tests found (.test., .spec. or tests/ files)
• expected: PASS: the mocha convention test/*.js run by npm test on 29 Node versions. The same run also credits
  nyc for coverage (PASS "branch coverage is reported with nyc"), which wraps npm test; a project with measured
  coverage and "no tests" is self-contradictory.
• evidence: test/index.js:1-4: var assert = require('assert') ...; package.json:
  "test": "mocha --reporter spec --check-leaks test/", "test-ci": "nyc --reporter=lcovonly --reporter=text npm test";
  .github/workflows/ci.yml "Run tests" step: npm run test-ci / npm test; validator
  project_validator/ecosystems/javascript/__init__.py:309: path=re.compile(r"(?:^|/)(?:[^/]*(?:[._]test|\.spec)\.[^/]+|__tests__/.+)$").
• repro: test/index.js with describe(/it( + "test": "mocha test/" + a step run: npm test.
• confidence: high

actions/checkout / tests-run

• kind: FALSE_NEGATIVE
• actual: FAIL "JavaScript/TypeScript: tests found in 10 file(s), but no CI configuration runs jest, vitest, mocha,
  node --test, deno test, bun test or playwright (or npm test)".
• expected: PASS: CI runs npm test, whose script is Jest's documented ESM invocation.
• evidence: .github/workflows/test.yml:27: - run: npm test; package.json:12: "test": "node --experimental-vm-modules node_modules/jest/bin/jest.js". Validator: ecosystems/javascript/__init__.py:314-321:
  _js_tests accepts name in _JS_RUNNERS or (name == "node" and "--test" in argv); node <path>/jest.js matches
  neither, and npm test itself (named in the hint) is not credited when its script is unrecognised.
• repro: package.json "test": "node --experimental-vm-modules node_modules/jest/bin/jest.js", src/a.test.ts,
  workflow run: npm test.
• confidence: high

octokit/graphql-schema / tests-run

• kind: FALSE_NEGATIVE
• actual: FAIL: no tests found (*.test.*, *.spec.* or __tests__/ files)
• expected: PASS: a test/ directory of node:assert tests runs on every push via npm test.
• evidence: package.json:13: "test": "node test",; test/index.js:1-2: import "./smoke.js"; / import "./validate.js";; test/smoke.js:1-5: import { strict } from "assert"; ... strict.equal(typeof validate, "function");; .github/workflows/test.yml:20: - run: npm test
• evidence (validator): ecosystems/javascript/init.py:307-309: name="*.test.*, *.spec.* or __tests__/ files" / path=re.compile(r"(?:^|/)(?:[^/]*(?:[._]test|\.spec)\.[^/]+|__tests__/.+)$"): a top-level test/ directory (mocha's and node's classic layout) is not a test location. npm test itself is accepted as a runner, so only the file pattern is missing.
• repro: package.json with "test": "node test", test/index.js, workflow run: npm test.
• confidence: high

expressjs/express, sindresorhus/got / tests-run

• kind: FALSE_NEGATIVE
• actual: FAIL: no tests found (*.test.*, *.spec.* or __tests__/ files) (both projects)
• expected: PASS: both have a test/ directory of test files that CI runs through a known runner.
• evidence:
  • express package.json:94: "test": "mocha --require test/support/env --reporter spec --check-leaks test/ test/acceptance/";
    package.json:95: "test-ci": "nyc --exclude examples --exclude test --exclude benchmarks --reporter=lcovonly --reporter=text npm test";
    .github/workflows/ci.yml:80: run: npm run test-ci; test/ holds 74 files (test/app.js, test/Router.js, ...)
  • got package.json: "test": "xo && tsc --noEmit && NODE_OPTIONS='--import=tsx/esm' node --test --test-timeout=60000 'test/*.ts'";
    .github/workflows/main.yml:26: - run: npm run test; test/ holds 43 .ts files (test/abort.ts, test/http.ts, ...)
  • validator: project_validator/ecosystems/javascript/__init__.py:309:
    path=re.compile(r"(?:^|/)(?:[^/]*(?:[._]test|\.spec)\.[^/]+|__tests__/.+)$") — nothing under test/ or tests/
    counts unless the file name itself says test. facts/roles.py:27 already maps test, tests, spec directories to
    the test role; the tests-run question does not use it. Note express's coverage check PASSes "branch coverage is
    reported with nyc" in the same run: the validator followed npm test to mocha for coverage but reports no tests (MESSAGE).
• repro: package.json {"scripts":{"test":"mocha test/"}}, test/app.js with one describe, workflow step run: npm test.
• confidence: high

fastify/fastify / tests-run, coverage

• kind: UNSUPPORTED
• actual: FAIL: tests found in 198 file(s), but no CI configuration runs jest, vitest, mocha, node --test, deno test, bun test or playwright (or npm test); coverage FAIL no branch coverage measurement found
• expected: PASS: CI runs npm run unit -> borp (a node:test runner wrapping c8) and npm run coverage:ci-check-coverage -> borp --coverage --check-coverage --lines 100.
• evidence: package.json:29: "unit": "borp"; package.json:16: "coverage:ci-check-coverage": "borp --reporter=@jsumners/line-reporter --coverage --check-coverage --lines 100";
  .github/workflows/ci.yml:85-87: - name: Run tests / run: | / npm run unit; .github/workflows/coverage-nix.yml (called from ci.yml:45): run: | / npm run coverage:ci-check-coverage;
  .borp.yaml:1-3: files: / - 'test/**/*.test.js'. Validator ecosystems/javascript/__init__.py:311: _JS_RUNNERS = frozenset({"jest", "vitest", "mocha", "ava"}).
  borp is used across the fastify organisation (and tap/node-tap is similarly absent). test:types -> tstyche (type tests) is also unknown to type-checking.
• repro: package.json {"scripts":{"unit":"borp"}} + .borp.yaml + step run: npm run unit.
• confidence: high (as a gap)

sindresorhus/got / static-analysis, compiler-warnings, warnings-as-errors

• kind: UNSUPPORTED
• actual: FAIL: no static analyzer runs; FAIL no JavaScript linters found; enable at least eslint
• expected: PASS: xo (the eslint wrapper used by every sindresorhus package and thousands of others) runs in CI through npm run test, configured in package.json's xo key.
• evidence: package.json: "test": "xo && tsc --noEmit && ...", "xo": [ { "ignores": [...] }, { "rules": { "@typescript-eslint/no-empty-function": "off", ... } } ];
  .github/workflows/main.yml:26: - run: npm run test.
• repro: package.json {"scripts":{"test":"xo"},"xo":{}} + step run: npm test.
• confidence: high (as a gap)

colinhacks/zod / tests-run, static-analysis, code-formatting

• kind: UNSUPPORTED, MESSAGE
• actual: FAIL: tests found in 203 file(s), but no CI configuration runs jest, vitest, ...; FAIL biome is configured (biome.jsonc, ...) but does not run; WARN biome.jsonc ... is defined but CI does not run ... biome
• expected: PASS for all three: CI runs the package scripts through the nub runner: nub run test -> vitest run, nub run lint:check -> biome lint ., nub run format:check -> biome check ..
  Also: the strict and --allow-unknown outputs are identical (diff SAME), i.e. an unknown program nub produced a confident FAIL, not an "unknown".
• evidence: .github/workflows/test.yml:28: uses: nubjs/setup-nub@47e5e7393d50f4e9544ddaf756aa277972afba2d # v0; test.yml:37: - run: nub run test;
  test.yml:52-53: - run: nub run format:check / - run: nub run lint:check; package.json:78: "test": "nub exec --node vitest run";
  package.json:72: "format:check": "nub exec --node biome check ."; package.json:74: "lint:check": "nub exec --node biome lint .".
  Validator ecosystems/javascript/__init__.py:87-110 knows npm/pnpm/yarn/bun/deno only. nub is new and rare, so this is a gap, but
  <unknown> run <script> where <script> is a package script could be treated as a script runner, or at least as unread.
• repro: step run: nub run test with package.json {"scripts":{"test":"vitest run"}}.
• confidence: medium

sindresorhus/got / type-checking (strict hint)

• kind: MESSAGE
• actual: PASS: tsc (tsconfig.json) runs + hint consider enabling strict mode: "strict": true in the compilerOptions of tsconfig.json
• expected: the hint should allow for an extends base: got's tsconfig extends @sindresorhus/tsconfig, which is strict.
• evidence: tsconfig.json:2: "extends": "@sindresorhus/tsconfig",. The base lives in node_modules, so it cannot be read; saying "strict is not visible (extends @sindresorhus/tsconfig)" would be accurate.
• repro: tsconfig.json {"extends":"@tsconfig/strictest"} (the docs say @tsconfig/strictest is recognised, so a named-base allowlist exists and could grow).
• confidence: low

---

Found by the regression-corpus audit (235 real repositories, one or more per ecosystem profile; corpus and expected results in corpus/). Evidence is quoted from the audited checkouts at the commits pinned in corpus/projects.toml; <checkouts>/<owner>__<repo> is such a checkout.

🤖 Generated with Claude Code

https://claude.ai/code/session_01ByTXNqDWdM7HzezrFtEvQt