test(contract): record smoke cassette to enable test_user_info_smoke

[millsmillsymills]

This was generated by AI during triage.

Summary

Task 10 of the PR A plan (docs/superpowers/plans/2026-05-12-live-contract-tests-pr-a-scaffold.md) is maintainer-only — requires a real sandbox PAT scoped to teamrocket.network. Once the cassette is recorded and committed, the @pytest.mark.skipif on test_user_info_smoke becomes inert and the test runs in CI.

Why this is ready-for-human

Requires a Gandi sandbox PAT. Cannot be done by an agent.

Steps for the maintainer

1. Create a sandbox PAT at https://account.gandi.net/, scoped to teamrocket.network only.
2. Store it: pass insert gandi/pat-sandbox (or equivalent).
3. From the repo root:

   GANDI_TOKEN=$(pass show gandi/pat-sandbox) make refresh-cassettes

4. Review the cassette diff for unredacted PII per CONTRIBUTING.md "Recording contract cassettes".
5. Commit tests/contract/cassettes/test_user_info_smoke.yaml.
6. Remove the @pytest.mark.skipif decorator from tests/contract/test_smoke.py::test_user_info_smoke (the cassette now exists, the skip is no longer load-bearing).
7. Open a PR with the cassette + the skipif removal.

Acceptance

• Cassette committed with PII redacted.
• Skipif decorator removed.
• CI passes test_user_info_smoke on replay-only.

[millsmillsymills]

🤖 Automated multi-agent issue review

Recommendation: keep-open · Priority: P3 · Effort: small · Premise still valid: True

Review. Valid, current, and precisely specified maintainer task. The smoke cassette has never been recorded (only .gitkeep present), the skipif on test_user_info_smoke remains in place, and conftest's record_mode='none' makes the skip load-bearing. It is correctly labeled ready-for-human: recording requires a real Gandi sandbox PAT scoped to teamrocket.network, which no agent can obtain. Sibling tasks #101 (PR B) and #102 (PR C) are the equivalent bulk cassette-recording work and remain open.

Suggested next step. Keep open as a maintainer-only task. The human should create a sandbox PAT, run GANDI_TOKEN=$(pass show gandi/pat-sandbox) make refresh-cassettes, commit tests/contract/cassettes/test_user_info_smoke.yaml (PII-redacted), remove the skipif at test_smoke.py:27, and open a PR. Low priority since the rest of the contract tier (#101/#102) is also pending; consider folding this single-cassette task into the #101 PR B effort to avoid a near-duplicate one-line PR.

Adversarial re-review (concurs). premise recheck: confirmed

The review is correct on all load-bearing points: the premise reproduces exactly against main, the issue is valid, current, well-specified, genuinely human-only (needs a real sandbox PAT), and keep-open at P3/small is right. The only defensible objection is the secondary 'fold into #101' suggestion, which contradicts the smoke test's documented purpose of validating the pipeline before the bulk PR B/C recording; #111 should stand alone and land first. That does not change the recommendation.

Open points raised:

• Suggested 'fold this single-cassette task into the test(contract): PR B — record cassettes for all 34 read endpoints #101 PR B effort to avoid a near-duplicate one-line PR' is the one weak spot. test(contract): PR B — record cassettes for all 34 read endpoints #101 is explicitly NOT execution-ready ('PR B does not yet have an implementation plan'; blocked on two human-judgment items per its own body), whereas test(contract): record smoke cassette to enable test_user_info_smoke #111 is a fully-specified, mechanical run-one-command-and-commit task. More importantly, test_smoke.py's own docstring states the smoke cassette exists 'to prove the record + replay pipeline works end-to-end BEFORE PR B starts recording the full API surface.' Folding test(contract): record smoke cassette to enable test_user_info_smoke #111 into test(contract): PR B — record cassettes for all 34 read endpoints #101 inverts the documented sequencing the smoke test was designed to enforce. test(contract): record smoke cassette to enable test_user_info_smoke #111 should stand alone and ideally land first.
• Priority P3 is slightly conservative: this smoke cassette is the gate that validates the entire Tier-3 record/replay pipeline before the bulk test(contract): PR B — record cassettes for all 34 read endpoints #101/test(contract): PR C — record cassettes for all 29 non-purchasing write endpoints #102 recording work. It is low-urgency (nothing currently depends on it) but higher-leverage than a typical P3 single test. Defensible either way.

_{Posted by a review + adversarial-review agent sweep across all open issues. No code was changed; no PRs were opened.}