e2e/tests/prisma-example-readme.e2e.test.ts (1)

66-94: 💤 Low value

Remove unnecessary async keywords.

The functions snapshotTransientOutputs, restoreTransientOutputs, and wipeTransientOutputs are declared async but contain only synchronous operations (no await statements). The async keyword is unnecessary and can be removed along with the Promise<...> return types.

♻️ Simplify function signatures

-async function snapshotTransientOutputs(): Promise<string> {
+function snapshotTransientOutputs(): string {
   const snap = mkdtempSync(join(tmpdir(), 'prisma-readme-e2e-snap-'))
   // ...
 }

-async function restoreTransientOutputs(snap: string): Promise<void> {
+function restoreTransientOutputs(snap: string): void {
   for (const rel of TRANSIENT_PATHS) {
   // ...
 }

-async function wipeTransientOutputs(): Promise<void> {
+function wipeTransientOutputs(): void {
   for (const rel of TRANSIENT_PATHS) {
   // ...
 }

Then remove await from the call sites on lines 111-112, 170, if these changes are applied.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@e2e/tests/prisma-example-readme.e2e.test.ts` around lines 66 - 94, The three
functions snapshotTransientOutputs, restoreTransientOutputs, and
wipeTransientOutputs are declared async but use only synchronous fs operations;
remove the async keyword and change their signatures from
Promise<string>/Promise<void> to plain string/void (e.g., function
snapshotTransientOutputs(): string { ... }), and update all call sites that
currently await them (calls to snapshotTransientOutputs(),
restoreTransientOutputs(), wipeTransientOutputs()) to remove the unnecessary
await so they are invoked synchronously.

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

e2e/tests/prisma-example-readme.e2e.test.ts (2)

145-159: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Increase beforeAll timeout to match step-budget worst case.

Line 159 keeps 600_000 ms, but configured/default step timeouts can exceed that in slow CI paths, causing wrapper timeout before step-level failures are reported.

Suggested minimal fix

-  }, 600_000) // 10 min total budget for the cold path
+  }, 900_000) // 15 min budget to cover worst-case step timeout sum

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@e2e/tests/prisma-example-readme.e2e.test.ts` around lines 145 - 159, The
beforeAll wrapper timeout (currently the fixed 600_000) can be shorter than the
sum of per-step timeouts and causes premature test termination; replace the
hardcoded 600_000 with a computed total budget based on the README_COMMANDS step
timeouts (e.g. const totalBudget = README_COMMANDS.reduce((acc, line) => acc +
timeoutFor(line), 0) + 60_000) and use totalBudget as the beforeAll timeout so
the outer hook always exceeds the worst-case step budget when running runStep
over README_COMMANDS (skip logic with SKIP_COMMANDS and outcomes map remains
unchanged).

---

33-34: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Redact process output before appending failure diagnostics.

Line 33 and Line 34 still include raw stderr/stdout, which can emit plaintext values in CI logs. Please sanitize (or summarize/truncate) these fields before lines.push(...).

Suggested minimal fix

+function redactPlaintext(s: string): string {
+  return s
+    .replace(/\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b/gi, '[redacted-email]')
+    .replace(/\b(?:postgres|mysql|mongodb):\/\/\S+/gi, '[redacted-conn-string]')
+    .slice(0, 4000)
+}
+
 function describeSpawnFailure(result: StepResult): string {
   const lines = [`step \`${result.label}\` failed.`]
   if (result.error) lines.push(`  spawn error: ${result.error.message}`)
   if (result.signal) lines.push(`  killed by signal: ${result.signal}`)
   if (typeof result.status === 'number') lines.push(`  exit status: ${result.status}`)
-  if (result.stderr.trim()) lines.push(`--- stderr ---\n${result.stderr.trim()}`)
-  if (result.stdout.trim()) lines.push(`--- stdout ---\n${result.stdout.trim()}`)
+  const safeStderr = redactPlaintext(result.stderr).trim()
+  const safeStdout = redactPlaintext(result.stdout).trim()
+  if (safeStderr) lines.push(`--- stderr ---\n${safeStderr}`)
+  if (safeStdout) lines.push(`--- stdout ---\n${safeStdout}`)
   return lines.join('\n')
 }

As per coding guidelines, Never log plaintext data; the library by design never logs plaintext sensitive information.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@e2e/tests/prisma-example-readme.e2e.test.ts` around lines 33 - 34, The test
currently appends raw process output (result.stderr and result.stdout) into
diagnostics via lines.push, which can leak plaintext; replace those direct
appends with sanitized or truncated summaries: for example compute a redacted
string from result.stderr and result.stdout (e.g., mask sensitive patterns,
limit length, or replace with "<redacted>"/first-N-chars + "...") and push that
sanitized string instead; update the lines.push calls that reference
result.stderr, result.stdout so they use the sanitized variables and include
contextual labels like "--- stderr ---" / "--- stdout ---" as before.