System Test Base Class & VCR Documentation

<objective> Enhance the existing ApplicationSystemTestCase with proper Capybara wait configuration, screenshot-on-failure, and temp file cleanup. Extract shared system test helpers (purge_solid_queue_tables, assert_item_order, etc.) from individual test files into a SystemTestHelpers module. Create VCR cassette maintenance documentation. (T1, T9, T16, T17) </objective> <context> @.claude/skills/sm-engine-test/SKILL.md -- test infrastructure patterns @.claude/skills/tdd-cycle/SKILL.md -- Minitest test conventions </context> <tasks> <task type="auto"> <name>Enhance ApplicationSystemTestCase</name> <files> test/application_system_test_case.rb </files> <action> Update test/application_system_test_case.rb to add: 1. Capybara configuration: `Capybara.default_max_wait_time = 5` (centralize the scattered wait: 5 values) 2. Common setup: `SourceMonitor.reset_configuration!` and `SourceMonitor::Jobs::Visibility.reset!` / `setup!` 3. Teardown with cleanup: reset config, reset Jobs::Visibility, call `clean_test_tmp_files` (conditionally clean test/tmp/ artifacts older than 1 hour) 4. Screenshot on failure: override `after_teardown` to save screenshot if test failed (Capybara's built-in `take_failed_screenshot` if available) 5. Keep existing driven_by, ActionCable::TestHelper, Turbo::Broadcastable::TestHelper includes

Do NOT add purge_solid_queue_tables to base setup -- only dashboard tests need it. That goes in SystemTestHelpers as an opt-in method. </action> <verify> ruby -c test/application_system_test_case.rb (syntax valid) </verify> <done> ApplicationSystemTestCase has Capybara wait config, setup/teardown, screenshot-on-failure, and tmp cleanup </done> </task> <task type="auto"> <name>Create SystemTestHelpers module</name> <files> test/support/system_test_helpers.rb </files> <action> Create test/support/system_test_helpers.rb with a SystemTestHelpers module containing helpers extracted from system test files:

1. purge_solid_queue_tables -- from dashboard_test.rb (purges SolidQueue tables)
2. seed_queue_activity(source:) -- from dashboard_test.rb (seeds queue data for testing)
3. apply_turbo_stream_messages(page) -- from dashboard_test.rb (processes turbo stream messages in Capybara)
4. parse_turbo_streams(html) -- from dashboard_test.rb (parses turbo stream HTML)
5. assert_item_order(expected_titles) -- from items_test.rb (asserts item display order)

Include this module in ApplicationSystemTestCase via require_relative + include. </action> <verify> ruby -c test/support/system_test_helpers.rb (syntax valid) </verify> <done> SystemTestHelpers module exists with all extracted helpers, included in ApplicationSystemTestCase </done> </task> <task type="auto"> <name>Refactor system tests to use shared base</name> <files> test/system/dashboard_test.rb test/system/items_test.rb test/system/logs_test.rb test/system/sources_test.rb test/system/mission_control_test.rb test/system/dropdown_fallback_test.rb </files> <action> For each system test file:

1. Remove local setup/teardown logic now handled by ApplicationSystemTestCase (config reset, Jobs::Visibility reset/setup)
2. Remove local helper method definitions that are now in SystemTestHelpers (purge_solid_queue_tables, assert_item_order, etc.)
3. Keep test-specific setup that is NOT shared (e.g., dashboard_test still calls purge_solid_queue_tables in its own setup since not all system tests need it, but the METHOD is defined centrally)
4. Remove explicit wait: 5 from assertions where Capybara's default_max_wait_time (now 5) handles it. Keep explicit waits only where they differ from default (e.g., wait: 10 for slow operations)
5. Verify each test file still requires "application_system_test_case" </action>

<verify> PARALLEL_WORKERS=1 bin/rails test test/system/ -- all system tests pass </verify> <done> System test files use shared base class setup/teardown and SystemTestHelpers. No duplicated helper definitions across system test files. </done> </task> <task type="auto"> <name>Create VCR cassette maintenance guide</name> <files> test/VCR_GUIDE.md </files> <action> Create test/VCR_GUIDE.md documenting: 1. **Overview**: VCR is configured in test_helper.rb, hooks into WebMock, cassettes stored in test/vcr_cassettes/ 2. **Naming convention**: `source_monitor/<domain>/<scenario>` (e.g., source_monitor/fetching/rss_success) 3. **When to use VCR vs WebMock stubs**: VCR for complex multi-request flows with real response bodies; WebMock stubs for simple single-request mocks where response content doesn't matter 4. **Recording new cassettes**: `VCR.use_cassette("name", record: :new_episodes) { ... }` then commit the YAML 5. **Regenerating stale cassettes**: Delete the YAML file, run the test with `record: :new_episodes`, review diff 6. **Current cassettes**: List the 4 existing cassettes and their purpose 7. **Maintenance**: Check cassettes annually or when upstream feed formats change; cassettes with expired SSL certs need re-recording </action> <verify> test -f test/VCR_GUIDE.md (file exists) </verify> <done> VCR_GUIDE.md exists with naming convention, recording strategy, regeneration guide, and cassette inventory </done> </task> </tasks> <verification> 1. PARALLEL_WORKERS=1 bin/rails test test/system/ -- all system tests pass 2. bin/rails test -- full suite passes 3. grep -r "def purge_solid_queue_tables" test/ returns only test/support/system_test_helpers.rb (plus test/lib/ if that has its own copy for non-system tests) 4. grep "default_max_wait_time" test/application_system_test_case.rb returns a match 5. test -f test/VCR_GUIDE.md </verification> <success_criteria> - ApplicationSystemTestCase has Capybara.default_max_wait_time = 5, screenshot-on-failure, and tmp cleanup - SystemTestHelpers module centralizes shared helpers from system test files - System test files no longer define their own copies of shared helpers - test/VCR_GUIDE.md documents cassette naming, recording, and maintenance strategy - All system tests pass without regressions </success_criteria> <output> 02-SUMMARY.md </output>