Garry's Opinionated OpenClaw/Hermes Agent Brain
TypeScript
/**
* #3927 — check-wasm-embedded.sh could false-fail under pipefail.
*
* `if ! echo "$OUTPUT" | grep -q '<pattern>'; then FAIL` runs under
* `set -euo pipefail`. grep -q exits as soon as it finds a match and
* can close its end of the pipe before echo finishes writing $OUTPUT;
* echo then gets SIGPIPE and the pipeline's exit status goes non-zero
* even though grep matched. Reproduced 50/50 by the issue reporter with
* a ~300KB payload.
*
* The fix is a plain bash substring test instead of a pipe, so this is
* a source-shape guard rather than an execution test: exercising the
* real script means a `bun build --compile` + binary run per check:wasm
* invocation, which is already covered (slowly) by `bun run check:all`
* and isn't worth duplicating here. This test just pins that the
* vulnerable pipe shape doesn't come back.
*/
import { describe, it, expect } from 'bun:test';
import { readFileSync } from 'node:fs';
import { resolve } from 'node:path';
const REPO_ROOT = resolve(import.meta.dir, '..', '..');
const SCRIPT = resolve(REPO_ROOT, 'scripts', 'check-wasm-embedded.sh');
describe('check-wasm-embedded.sh (#3927)', () => {
const source = readFileSync(SCRIPT, 'utf8');
it('still runs under set -euo pipefail', () => {
expect(source).toContain('set -euo pipefail');
});
it('does not pipe $OUTPUT through grep -q', () => {
// Match the vulnerable pipeline shape in live code (ignore commented examples).
expect(source).not.toMatch(/^[^\n#]*echo "\$OUTPUT"\s*\|\s*grep\s+-q\b/m);
});
it('uses a plain bash substring test for each of the three checks', () => {
expect(source).toContain(`if [[ "$OUTPUT" != *'"has_symbol_names": true'* ]]; then`);
expect(source).toContain(`if [[ "$OUTPUT" != *'"has_typescript_header": true'* ]]; then`);
expect(source).toContain(`if [[ "$OUTPUT" != *'"calculateScore"'* ]]; then`);
});
});