moj-projekt

Market Intelligence Platform - evidence-backed market narrative intelligence. See docs/README.md for the full documentation set.

The operator's path from clone to a stored Document, including PowerShell vs bash, is docs/reference/WORKFLOWS.md. Do not treat this README as a second copy of those commands.

Quickstart

Python 3.11+, Docker Desktop, then:

1. python -m venv .venv and activate (.venv\Scripts\Activate.ps1 on Windows, source .venv/bin/activate on macOS/Linux).
2. pip install -e ".[dev]"
3. pre-commit install --install-hooks and pre-commit install --hook-type pre-push
4. Copy .env.example to .env (Copy-Item on PowerShell, cp on bash). Host-side commands use localhost:5433; the app container still uses db:5432.
5. docker compose up -d then alembic upgrade head
6. python -m moj_projekt.persistence.seed_sources
7. python -m moj_projekt.cycle.run_once

GET http://localhost:8000/health should report database: ok and pgvector: available. The cycle writes Documents from the live Bloomberg Markets RSS feed; Reuters/AP seed URLs fail in isolation and do not fail the cycle.

Checks

python scripts/check.py runs ruff check, strict mypy over src/, then the unit suite. Default pytest excludes integration.

Integration tests need the compose database already migrated:

python -m pytest -m "integration and not network"

Live RSS is marked network (and integration); it needs outbound HTTP and is excluded from CI. See WORKFLOWS.md for the exact commands.

CI (.github/workflows/ci.yml) runs lint, type-check, and unit + integration (pytest -m "not network") on every push/PR into main and sprint/**. Local pre-commit mirrors the DB-free subset: commit gets hygiene + ruff + mypy; push gets the unit suite.

Status

Sprint 001 ("Foundation to first real document") is in progress. See docs/planning/CURRENT_STATUS.md for what is implemented today and docs/planning/sprints/SPRINT_001.md for the full task breakdown.

More

Full documentation: docs/README.md.