Jobber/documentation/self-hosting.md
Marcin Szymański cc84157149
support RXResume config URL config in onboarding (#159)
* support RXResume config URL config in onboarding

* update documentation and fix linting
2026-02-12 22:04:32 +00:00

137 lines
4.9 KiB
Markdown

# Self-Hosting (Docker Compose)
The easiest way to run JobOps is via Docker Compose. The app is self-configuring and will guide you through the setup on your first visit.
## Prereqs
- Docker Desktop or Docker Engine + Compose v2
## 1) Start the stack
No environment variables are strictly required to start. Simply run:
```bash
docker compose up -d
```
This pulls the pre-built image from **GitHub Container Registry (GHCR)** and starts the API, UI, and scrapers in a single container. The image is multi-arch (supports `amd64` and `arm64`), making it compatible with Apple Silicon and Raspberry Pi.
If you want to build it yourself, you can run `docker compose up -d --build`.
## 2) Access the app and Onboard
Open your browser to:
- **Dashboard**: http://localhost:3005
On first launch, you will be greeted by an **Onboarding Wizard**. The app will help you validate and save your configuration:
1. **LLM Provider**: OpenRouter is the default. Add an API key if required (OpenRouter/OpenAI/Gemini), or configure a local base URL (LM Studio/Ollama).
2. **PDF Export**: Add your RxResume credentials (used to export PDFs from v4.rxresu.me).
3. **Template Resume**: Select a base resume from your v4.rxresu.me account.
The app saves these to its persistent database, so you don't need to manage `.env` files for basic setup. All other settings (like search terms, job sources, and more) can also be configured directly in the UI.
Upgrade note: `OPENROUTER_API_KEY` is deprecated. Existing OpenRouter keys are automatically migrated/copied to `LLM_API_KEY` so you don't lose them.
## Gmail OAuth (Post-Application Inbox)
If you want to connect Gmail in the Tracking Inbox page, configure Google OAuth credentials for the API server.
### 1) Create Google OAuth credentials
In Google Cloud:
1. Open your project (or create one), then configure the OAuth consent screen.
2. Enable the Gmail API.
3. Create an OAuth client ID (`Web application` type).
4. Add an authorized redirect URI:
- `http://localhost:3005/oauth/gmail/callback` (default local setup)
- or your deployed app URL, for example `https://your-domain.com/oauth/gmail/callback`
### 2) Configure environment variables
Set these on the JobOps container:
- `GMAIL_OAUTH_CLIENT_ID` (required)
- `GMAIL_OAUTH_CLIENT_SECRET` (required)
- `GMAIL_OAUTH_REDIRECT_URI` (optional, recommended for production)
- If omitted, JobOps derives it from the incoming request host as `/oauth/gmail/callback`.
### 3) Restart and connect
After setting env vars, restart the container and use `Tracking Inbox -> Connect Gmail`.
Notes:
- JobOps requests `gmail.readonly` scope.
- If Google returns no refresh token, disconnect and re-connect to force a fresh consent flow.
## Email-to-Job Matching Overview
When Gmail sync runs, your emails are automatically analyzed and routed. Here's what happens:
```mermaid
flowchart TD
A[Recruitment email arrives in Gmail] --> B[Smart Router AI analyzes content]
B --> C{How confident is the match?}
C -->|95-100%| D[Auto-linked to job]
D --> E[Timeline updated automatically]
C -->|50-94%| F[Goes to Inbox for review<br/>with suggested job match]
C -->|<50%| G{Is it relevant?}
G -->|Yes| H[Goes to Inbox as orphan<br/>relevant but job unclear]
G -->|No| I[Ignored - not job-related]
F --> J{You review in Inbox}
H --> J
J -->|Approve| K[Linked to selected job<br/>Timeline updated]
J -->|Ignore| L[Marked as not relevant]
```
**What the AI looks for:**
- Content relevance - Is this about job applications?
- Job matching - Which of your tracked jobs is this about?
- Message type - Interview, offer, rejection, or update?
**Your control:**
- High-confidence matches (95%+) happen automatically
- Everything else appears in your Inbox for a quick yes/no decision
- You can always correct the job match when approving
**Privacy note:** Only job ID, company name, and title are sent to the AI for matching. Full email content stays local.
## Persistent data
`./data` is bind-mounted into the container. It stores:
- SQLite DB: `data/jobs.db` (contains your API keys and configuration)
- Generated PDFs: `data/pdfs/`
- Template resume selection: Stored internally after selection.
## Public demo deployment (`DEMO_MODE=true`)
For a public sandbox website, set `DEMO_MODE=true` on the container.
Behavior in demo mode:
- **Works (local demo DB):** browsing, filtering, job status updates, timeline edits.
- **Simulated (no external side effects):** pipeline run, job summarize/process/rescore/pdf/apply, onboarding validations.
- **Blocked:** settings writes, database clear, backup create/delete, status bulk deletes.
- **Auto-reset:** seeded demo data is reset every 6 hours.
## Updating
```bash
git pull
docker compose pull
docker compose up -d
```
## Self-hosted Reactive Resume
If you are also self-hosting Reactive Resume, you should configure the url
by setting `RXRESUME_URL` to your instance (e.g. `http://rxresume.local.net`).