Enable MistDemo integration workflow on claude/** branches

[leogdion]

Summary

• Adds claude/** to the push trigger of .github/workflows/MistDemo-Integration.yml so the live CloudKit integration workflow runs on working branches, letting us watch it pass while iterating.
• Branch is synced to the latest v1.0.0-beta.2 (scaffold Scaffold MistDemo (CLI + App + Web) for v1.0.0-beta.2 endpoints #371 + the Note record-type fix).

Status

The integration workflow already ran green on this branch — run 26251422359 (Success): live query/create/modify/cleanup against the Note type, demo-errors (401/404/409), and demo-in-filter all passed.

Note

The claude/** trigger is a temporary enablement for iteration and should be dropped before this merges back into v1.0.0-beta.2.

https://claude.ai/code/session_01EvLrWZwcSs1MjiCrUx8KjU

---

Generated by Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: e4e17c08-c559-4f4c-a1d0-04c1b2141a14

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch claude/zealous-lovelace-VnPSo

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[claude]

Code Review

Summary

This is a minimal, well-scoped CI configuration change — one line added to the push trigger. The PR description is clear and honest about intent. Here are my observations:

---

What works well

• Minimal footprint: The diff is exactly the right size for the goal; no unnecessary changes.
• Self-documenting PR: The description clearly explains why this exists, that the integration run already passed, and that the trigger must be removed before merging upstream. That transparency is the right pattern for a temporary enablement.
• No secrets exposure risk from forks: The workflow already has no pull_request trigger, so secrets remain confined to pushes from repo collaborators, which applies equally to claude/**.

---

Concerns

1. Acknowledged but unguarded technical debt

The PR notes the claude/** trigger "should be dropped before this merges back into v1.0.0-beta.2." That's the right intent, but there's nothing in the workflow itself or branch protection that enforces the removal. A TODO comment directly in the YAML would make the intent visible to anyone who edits the workflow later and prevent it from silently becoming permanent:

      - 'claude/**'   # TODO: temporary – remove before merging to v1.0.0-beta.2

2. cancel-in-progress: false amplified by the new trigger pattern

The concurrency block reads:

concurrency:
  group: mistdemo-integration-${{ github.ref }}
  cancel-in-progress: false

For main and versioned release branches, cancel-in-progress: false is appropriate — you want every commit tested. For claude/** branches, where iteration is fast (many pushes in a short window), this means queued runs rather than superseded ones. Each full run is up to 75 minutes (45-min build + 30-min integration) and consumes CloudKit API quota. Consider overriding for the development pattern:

  cancel-in-progress: ${{ !startsWith(github.ref, 'refs/heads/v') && github.ref != 'refs/heads/main' }}

This cancels in-progress runs for claude/** and other non-release branches, preserving the safe behavior on main/release tags.

3. CI cost on noisy branches

Related to (2): a claude/** working branch can accumulate many pushes rapidly (rebases, fixups, automated commits). Without cancellation, each push queues a full build + integration run. If CI minutes or CloudKit quota are a concern, it may be worth adding a per-branch rate limit or restricting the pattern further (e.g., claude/integration-* or only specific named branches) rather than the wide glob.

---

Verdict

The change is safe and does what it says. The two concrete suggestions worth acting on before any wider merge are:

1. Add an inline # TODO comment so the temporary nature is visible in the file itself.
2. Evaluate whether cancel-in-progress: false is the right setting for development branches (it's probably not).

No issues with security, correctness, or test coverage for a CI config change of this scope.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (v1.0.0-beta.2@1376c8c). Learn more about missing BASE report.

Additional details and impacted files

@@               Coverage Diff                @@
##             v1.0.0-beta.2     #374   +/-   ##
================================================
  Coverage                 ?   68.21%           
================================================
  Files                    ?      120           
  Lines                    ?     2825           
  Branches                 ?        0           
================================================
  Hits                     ?     1927           
  Misses                   ?      898           
  Partials                 ?        0           

Flag	Coverage Δ
spm	68.03% <ø> (?)
swift-6.1-jammy	68.17% <ø> (?)
swift-6.1-noble	67.96% <ø> (?)
swift-6.2-jammy	68.03% <ø> (?)
swift-6.2-noble	68.10% <ø> (?)
swift-6.3-jammy	67.96% <ø> (?)
swift-6.3-noble	68.03% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[claude]

Code Review

Summary

This PR has three concerns bundled together: (1) enabling the live integration workflow on claude/** branches, (2) fixing the Note record-type in integration tests, and (3) fixing the asset/location oneOf decoder ambiguity. The core fixes are sound, but the PR is currently in a build-breaking state that needs to be resolved before merge.

---

🔴 Critical — Package does not build

The last commit updates openapi.yaml to mark latitude, longitude, and recordName as required, but does not commit the regenerated Sources/MistKitOpenAPI/Types.swift. The commit message explicitly acknowledges this:

"NOTE: requires regenerating Sources/MistKitOpenAPI via ./Scripts/generate-openapi.sh — the committed generated client is not yet updated, so the package will not build until regeneration."

The generated types still have:

• LocationValue.latitude: Swift.Double? — but Location.init(latitude:) requires Double (non-optional)
• LocationValue.longitude: Swift.Double? — same
• ReferenceValue.recordName: Swift.String? — but Reference.init(recordName:) requires String (non-optional)

The new init(locationValue:) and init(referenceValue:) in FieldValue+Components.swift pass these optional values to non-optional initializers — type mismatch, compile error.

Required action before merge: run ./Scripts/generate-openapi.sh and commit the updated Sources/MistKitOpenAPI/Types.swift.

---

🟡 Design — Reference.recordName remains non-optional after fix

After regeneration, ReferenceValue.recordName will be non-optional in the generated types. The domain Reference.recordName: String (non-optional) matches — good. However, the old guard was actually a defensive layer catching malformed API responses that violated the schema. Marking the field required in the schema is the right fix (the guard becomes unreachable), but it's worth a comment in CLAUDE.md or the commit message confirming that CloudKit Web Services always provides recordName in practice (the Apple docs back this up).

---

🟡 Breaking public API

Removing ConversionError.locationMissingCoordinates and ConversionError.referenceMissingRecordName is a public API breaking change. Any downstream code exhaustively switching over ConversionError will fail to compile. Given this is still v1.0.0-beta.2, breaking changes are expected — but worth calling out in the changelog/release notes.

---

🟡 Test: two removal, one addition — coverage gap for reference

The old tests verified that init(referenceValue:) and RecordInfo.init(from:) threw on a nil recordName. Those cases now can't be constructed (correct — the schema now forbids it). The new regression test for asset→location misidentification is well-written and directly targets the bug.

However, consider adding a positive-path test: a ReferenceValue with a valid recordName decoding to .reference (analogous to the new asset test). That would give you symmetric regression coverage.

---

✅ What's right

• Root-cause fix is correct. Making latitude/longitude/recordName required in the OpenAPI schema is the right lever — it fixes the oneOf first-match-wins ambiguity without any Swift-side hacks.
• Simplified Swift code. Removing the guard/throw in both init(locationValue:) and init(referenceValue:) is a net improvement once the schema enforces non-nullability at the decode boundary.
• Regression test. The assetFieldValueConvertsToAsset test is clean, clearly documents the bug being fixed, and directly decodes from JSON to validate the full decode→convert pipeline.
• CI workflow change. Adding claude/** temporarily is a reasonable iteration aid and the PR description correctly flags it for removal before merging to v1.0.0-beta.2.

---

Action items before merge

1. Run ./Scripts/generate-openapi.sh and commit the updated Sources/MistKitOpenAPI/Types.swift — the package must build.
2. Verify CI passes on the updated branch (the integration run cited in the PR body predates commit 0033cb6).
3. Remove claude/** from .github/workflows/MistDemo-Integration.yml (already noted in PR description).