- Replace conftest.py with two autouse fixtures: reset_graph_factory_singleton
and reset_task_manager_singleton, plus in_memory_db fixture for tests
requiring a real DB
- Rewrite test_project_task_recovery.py to use ProjectManager + SQLite in-memory
DB instead of the removed Project dataclass (same guarantee: active_task_id
persists and defaults to None)
- Fix db.py init_db() to import db_models before Base.metadata.create_all() so
all ORM tables are registered and created on first startup
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace file-based JSON persistence with SQLAlchemy DB (ProjectModel/ProjectFileModel)
and StorageService for file content; remove Project dataclass; add 7 passing tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace in-memory dict-based TaskManager with a SQLAlchemy-backed implementation
using TaskModel. Tasks now survive process restarts. 6 new tests added and passing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds ?format=md|pdf query param to GET /api/report/<id>/download.
PDF is generated from Markdown via fitz.Story (PyMuPDF). Also fixes
create_app() to support dict config and skip JWT auth in TESTING mode.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When LLM_PROVIDER=gemini, LLMClient automatically uses the Google AI Studio
OpenAI-compatible endpoint instead of requiring manual base_url configuration.
An explicit base_url argument still takes precedence.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Ubuntu