test(fixtures): restore apps-installed fixture with assertions

[pepmach]

Summary

Restores the apps-installed seed fixture, which seeds a KIROCREW_HOME containing one
locally installed app rather than a bare home:

• config.json — an onboarded home (default workspace default -> workspace, dark theme,
  host 127.0.0.1, UTC).
• apps/fixture-notes/installed.json — the persisted install record: enabled: true,
  origin: local, resources: gateway, lifecycle: gateway.
• apps/fixture-notes/app.json — a valid app manifest (name, displayName, version,
  description, author).
• apps/fixture-notes/data/notes.md — app-scoped data, so the data-dir resolver has a real
  file to find.
• fixture.yaml — fixture metadata at schema-version: 2026-04-28.

The point of the fixture is to give app-manager consumers a non-builtin subject: an
enabled local app with gateway-managed resources and gateway lifecycle metadata.

Consumer test

test/test_fixture_apps_installed.py drives the production readers only — it never re-reads
the fixture files it seeded, so the fixture is validated through the code path real callers
use:

• kiro_crew.apps.manager.list_apps() — asserts fixture-notes is listed with
  enabled is True, origin == "local", resources == "gateway",
  lifecycle == "gateway", and a parsed manifest["name"] == "fixture-notes".
• kiro_crew.apps.manager.app_data_dir("fixture-notes") — asserts the resolver locates the
  seeded notes.md.

Both run inside kiro_crew.testing.fixtures.seeded_home("apps-installed"), so a schema the
production reader rejects fails the test instead of passing silently.

Schema repairs

The install record drifted from what the app manager now persists. Repaired against the
preserved pre-split copy:

• installed_at renamed to installedAt (camelCase, matching the persisted key).
• Four newer persisted fields added: displayName, schemaVersion, dev,
  defaultOnBackfilled.

Everything else in the pre-split copy is preserved as-is.

Verification

.venv/bin/python -m pytest test/test_fixture_apps_installed.py test/test_pod_scenarios_command.py test/test_pod_seed_scenarios.py test/test_seed.py -q

125 passed, 0 failed.

The acceptance criterion is that this passes with the fixture present: main already
carries the registry-tolerant scenarios test from #8222, and the scenario/seed suites
enumerate the fixture registry, so a malformed or unregistered fixture would surface there
rather than only in its own test.

Style gates on the touched test file, all clean: isort --check-only, flake8,
black --check --target-version py310.

Pattern harvest

Not generalizable: fixture restoration validated against current production readers; no new rule.

Backend-only change, no screenshots: the diff adds test fixture data under
src/kiro_crew/tests_fixtures/apps-installed/ plus one test file under test/. No dashboard,
CLI-output, or other user-visible surface is touched.

[github-actions]

Design Review (Fable 5) — ✅ PASS

Design-level review of d096bf965f230f4591214f46ccb8dcef2242ea93 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

The verification checks out: the fixture's installed.json fields match the manager's current InstalledApp dataclass (installedAt, schemaVersion: 2, dev, defaultOnBackfilled), the fixture registry resolves by directory name so no registration edit is needed, packaging globs (setup.cfg, MANIFEST.in) already include tests_fixtures/**, and the test consumes only production readers (list_apps, app_data_dir) inside seeded_home, matching the pattern the fixtures helper was designed for.

Design-Verdict: PASS

Fixture is validated through the production readers it exists to serve, with schema fields verified against the manager's current persisted record.

[DESIGN-REVIEWED] d096bf9

[github-actions]

Opus 4.8 Review — ✅ no blocking findings

Reviewed d096bf965f230f4591214f46ccb8dcef2242ea93 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] d096bf9

Verdict parsed from the review's SHA-scoped output markers for commit d096bf965f230f4591214f46ccb8dcef2242ea93.

False positive or not applicable? A repository writer can comment:
/ai-review override fable d096bf965f230f4591214f46ccb8dcef2242ea93: <one-sentence reason>

[github-actions]

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of d096bf965f230f4591214f46ccb8dcef2242ea93 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] d096bf9

False positive or not applicable? A repository writer can comment:
/ai-review override gpt d096bf965f230f4591214f46ccb8dcef2242ea93: <one-sentence reason>

[github-actions]

First Principles Review (Fable 5) — ✅ PASS

Premise-level review of d096bf965f230f4591214f46ccb8dcef2242ea93 — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

All checks are done. The fixture's schema fields match what manager.py persists (installedAt, schemaVersion: 2, dev, defaultOnBackfilled at src/kiro_crew/apps/manager.py:117-138), the registry-driven pod suite auto-covers any new fixture, and no doc enumerates fixture names, so nothing else needed updating. Emitting the review.

First-Principles-Verdict: PASS

A data-only seed scenario the registry mechanism was built to absorb, validated through the production readers instead of by re-reading itself.

What this change ships

Intent: give operators and app-manager tests a seedable home containing one installed non-builtin app — a FIX by the author's framing (restoring a pre-split fixture), functionally an ADDITION of one scenario.

1. kirocrew pod scenarios / pod up --seed gains an apps-installed row, auto-discovered — justified
2. Scenario seeds one enabled local app: manifest, persisted install record, data file — justified; fields match manager.py's persisted schema
3. Scenario's home arrives onboarded (config.json with workspace, host, theme) — declared, skips onboarding in a seeded pod
4. New test drives manager.list_apps() / app_data_dir() against the seeded home — justified, the only reader-path validation the fixture has
5. Five small files join the runtime wheel as package data — inherent to the shipped-fixture mechanism seed.py documents

Every item is declared in the description; the registry suite (test/test_pod_seed_scenarios.py _ALL_FIXTURES) automatically extends manifest, seedability, JSON-parse, credential, and size gates to the new fixture, exactly as the description claims.

Watch

The byte-for-byte seed-fidelity contract is parametrized over hardcoded names, not the registry: 5 literal lists in test/test_fixtures_helper.py (grepped ["empty", "minimal", "rich"] — lines 34, 42, 160, 171, 190), so apps-installed sits outside the copytree-fidelity check while the pod suite already shows the registry-driven spelling. Replacing the two universal lists (34, 42) with the existing available_fixtures() registry would delete the divergence this fixture just created.

[FIRST-PRINCIPLES-REVIEWED] d096bf9

[chenmingwei23]

Tier 1 auto-approve: test (6 files). Criteria: no conflict, no requested changes, security path denylist clean, design-doc gate clean, SAST annotations clean, security checklist all-NO, AI reviewers green. Category: test(fixtures) restore apps-installed fixture data plus one test assertion file, no runtime code.

[dwu96]

Tier 1 auto-approve: test (6 files). Criteria: no conflict, no requested changes, security path denylist clean, design-doc gate clean, SAST annotations clean, security checklist all-NO, AI reviewers green. Category: purely additive test fixture — new tests_fixtures/apps-installed data tree plus one new test/ consumer test; no existing file modified and no runtime code path changed.