Fuzz testing (nightly) #1

	name: Fuzz testing (nightly)

	# High-iteration fuzz testing for the four binary format parsers
	# (EDF/BDF, BrainVision, EEGLAB, FIFF). 10,000 rounds per target on
	# corpus-seeded mutations of real fixture files — too slow for per-PR
	# gating, so it runs once per night.
	#
	# Per-PR confidence comes from `tests/prop-*.test.mjs` (100 runs each,
	# part of the normal `npm test` suite). Nightly fuzz extends that
	# depth by ~100× to surface rare crashes the small property tests
	# can't reach.
	#
	# Workflow:
	# - Scheduled at 04:00 UTC (after mutation job at 03:00, so they
	# don't compete for the same shared runner pool)
	# - Manual trigger via Actions → "Fuzz testing (nightly)" → Run
	# - Evidence artifacts uploaded on every run; the `tests/evidence/
	# fuzz/` directory may be empty when no finding has landed yet —
	# that's fine, the upload step degrades to a no-op.
	#
	# When a fuzz target finds a crash, the runbook is documented in
	# docs/qa-strategy.md § Fuzz Testing.

	on:
	schedule:
	- cron: '0 4 * * *' # 04:00 UTC nightly
	workflow_dispatch: {}

	env:
	FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: 'true'

	jobs:
	fuzz:
	runs-on: ubuntu-latest
	timeout-minutes: 30
	steps:
	- uses: actions/checkout@v4
	- uses: actions/setup-node@v4
	with:
	node-version: '20'
	cache: 'npm'
	- run: npm ci
	- name: Run fuzz suite
	run: npm run test:fuzz
	- name: Upload fuzz evidence
	if: always()
	uses: actions/upload-artifact@v4
	with:
	name: fuzz-evidence
	path: tests/evidence/fuzz/
	retention-days: 14
	if-no-files-found: ignore

Provide feedback