Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
import { realpathSync } from 'node:fs'
// Companion reporter: a successful file-process placeholder is not a test.
// Keep normal TAP output as the primary reporter and fail on missing coverage.
export default async function* registrationReporter(source) {
const requested = JSON.parse(process.env.NODE_TEST_EXPECTED_FILES || '[]')
if (!Array.isArray(requested) || requested.length === 0)
throw new Error('Expected explicit node test files')
const counts = new Map(requested.map(file => [realpathSync(file), 0]))
for await (const event of source) {
if (event.type !== 'test:pass') continue
const data = event.data
if (!data?.file || data.skip || data.todo || data.details?.type !== 'test') continue
const file = realpathSync(data.file)
if (!counts.has(file)) continue
// Empty files generate a synthetic case named after the input path.
let filePlaceholder = false
try {
filePlaceholder = realpathSync(data.name) === file
} catch {
/* Ordinary test name. */
}
if (!filePlaceholder) counts.set(file, counts.get(file) + 1)
}
const missing = [...counts].filter(([, count]) => count === 0).map(([file]) => file)
if (missing.length) throw new Error(`No executed test cases in: ${missing.join(', ')}`)
}