test(cli): pin bucket seed prompts (CLI-2433)

[7ttp]

TL;DR

fixes the stale claim that db reset bucket seeding takes confirmation defaults silently,
which was caused by the call site comment and
SIDE_EFFECTS.md never describing the real print and scan contract and is now fixed by correcting both and pinning the behavior with four integration tests covering the overwrite and prune defaults, the SUPABASE_YES dotenv path, and the piped answer path...

ref:

• closes: supabase db reset prompts on stderr and blocks on stdin during bucket seeding #6607

[github-actions]

🤖 AI Review

The PR changes only comments, documentation, and tests. Of Claude's five findings, two documentation/test-fidelity nits are confirmed and three are refuted by code scope, existing coverage, or trusted repository conventions. Codex reported no findings. Tests could not run because pr/node_modules is absent; verdicts are based on source inspection.

Findings

Severity	Location	Category	Sources	Claim
⚪ NIT	apps/cli/src/commands/db/reset/SIDE_EFFECTS.md:207	documentation	claude	The text-mode documentation omits that the stdin read also occurs on a TTY and is bounded to 100 ms.
⚪ NIT	apps/cli/src/commands/db/reset/reset.integration.test.ts:426	test-coverage	claude	The piped-input test does not model non-TTY stdin even though its fixture option is documented as non-TTY input.

Refuted findings (kept for transparency, not posted as review comments)

• apps/cli/src/commands/db/reset/reset.integration.test.ts:500 (documentation): The comment above mockStdin incorrectly says stdin is only a service dependency despite local bucket confirmations consuming pipedStdin.
  Refuted: The finding drops the comment's explicit remote-reset scope. For that path, TTY mode routes the confirmation through mockOutput, and stdin is only present to satisfy the Effect dependency. Local bucket tests use stdin separately.
• apps/cli/src/commands/db/reset/SIDE_EFFECTS.md:206 (test-coverage): The machine-format bucket-confirmation behavior needs another reset integration test because none of the four new tests use a machine format.
  Refuted: The underlying machine-format confirmation behavior and the local-reset JSON path already have integration coverage. Trusted/CLAUDE.md:171-172 explicitly prefers some test duplication avoidance and directs reviewers to reject redundant assertions.
• apps/cli/src/commands/db/reset/reset.integration.test.ts:463 (maintainability): The inline recording HTTP client should be extracted because similar route-matching mocks exist elsewhere.
  Refuted: There is no reusable recording-client factory being duplicated; the shared helper exposes a full storage-command setup with different routing capabilities. Trusted/CLAUDE.md:171 says to accept some test duplication rather than introduce unnecessary abstractions, and trusted/apps/cli/CLAUDE.md:35-39 says not to extract every helper used twice.

Stats

Claude findings: 5 · Codex findings: 0 · Confirmed: 2 · Refuted: 3 · Uncertain: 0

---

Models: claude-opus-5 + gpt-5.6-sol · Trigger: auto · Workflow run

This review runs once per PR. A maintainer can request another with a /ai-review comment.