Shared Test Fixtures and Coverage Gaps — pyrite

Summary

The test suite (456 tests) has good coverage of happy paths but significant gaps in error paths, search modes, and CLI commands. Test fixture setup is duplicated across 5 files.

Issues (from test evaluation)

Fixture Duplication

KB + config + DB + index setup is copy-pasted in:

test_rest_api.py, test_mcp_server.py, test_agent_cli.py, test_starred_entries.py, test_integration.py

Fix: Extract shared `pyrite_test_env` fixture into `conftest.py`.

Coverage Gaps (HIGH priority)

Endpoint error paths (400/404/500 responses) — barely tested

Semantic and hybrid search modes — untested

Query expansion (`expand=True`) — untested

24 of 42 CLI functions untested

MCP tier restrictions (read vs write vs admin) — untested

Plugin validation conflicts — untested

Coverage Gaps (MEDIUM priority)

Config validation edge cases (duplicate KB names, invalid YAML)

Migration rollback testing

Concurrent update scenarios

Template API endpoints (only service tested, not endpoints)

Test Quality

Many API tests only check status code, not response body structure

Integration-heavy: 83% integration tests, 17% unit tests

Consider `pytest -m integration` marker for slow tests

Acceptance Criteria

[x] Shared test fixture in conftest.py (composable: tmp_kb_dir, kb_configs, pyrite_config, pyrite_db, index_mgr, kb_service, sample_events, sample_person, indexed_test_env, rest_api_env)

[x] Error path tests for all endpoint modules (15 tests: 404, 400, empty results)

[ ] Semantic and hybrid search modes tested (deferred — requires embedding model setup)

[x] MCP tier restrictions tested (7 tests: read/write/admin tool exposure)

[x] 17+ new Typer CLI tests (list, get, timeline, tags, backlinks, create, update, delete, config)

[x] Pytest markers defined: integration, api, mcp, cli

Total: 495 tests (456 existing + 39 new)

Shared Test Fixtures and Coverage Gapsbacklog_item