# Session 2026-06-07 — Career MCP scaffold

## Decisions taken (from prior Claude.ai conversation, locked in this session)
- Project lives at `/home/m3ac/genoa-entwuerfe.com/jobscraper/` (renamed from `career-mcp/`)
- v1 sources: LinkedIn (public guest endpoint), Indeed (Playwright), Greenhouse, Lever
- Storage: MySQL `m3ac_jobscraper` on this VPS
- Interface: MCP primary, REST API secondary, worker for scheduled scrapes
- Three systemd services share one `.env`

## What got built
- Full file tree (50 files) matching the structure from the planning convo
- Real content: pyproject, .env.example, docker-compose, README, STRUCTURE.md, SPEC.md, db/models.py (full schema), schemas/*, config.py, logging.py, utils/http.py, alembic/env.py, api/main.py, mcp_server/server.py + tools.py (signatures), scripts/init_db.py + seed_sources.py, three systemd units
- Stubs (NotImplementedError) for: linkedin/indeed/greenhouse/lever scrapers, pipeline/* helpers, worker tasks

## Notes
- Caddyfile kept as reference; this box uses Apache so vhost config still needs writing for Phase 5
- SPEC.md (career-mcp-era) kept under docs/ — broader plan including auto-apply phase; supersedes nothing in the new structure but useful as long-range context

## Open before Phase 1 coding
1. Create MySQL `m3ac_jobscraper` DB + user, or wait for cPanel-side creation?
2. Proxy URL for LinkedIn requests, or accept bare-IP rate limiting in dev?