Quickstart
Meridian runs entirely on your machine. Clone the repository, copy the config templates, install
dependencies, and open the dashboard at http://127.0.0.1:3030.
Prerequisites
- Node.js 20 or later
- npm
- Chromium via Playwright (installed below)
- Optional: an OpenAI-compatible API key or local Ollama instance
1. Clone and configure
Fastest path: run the init script (copies config, installs deps, runs doctor).
git clone https://github.com/vijayptiwari/meridian.git
cd meridian
npm run init
Or configure manually:
git clone https://github.com/vijayptiwari/meridian.git
cd meridian
cp src/config.example.json src/config.json
cp .env.example .env
Video walkthrough
Record or embed a 3–5 minute install + demo run here. Suggested flow: npm run init → npm run ui → Try Demo.
2. Install dependencies
npm install
npx playwright install chromium
npm run doctor
npm run doctor verifies Node version, config files, and module integrity. Fix any reported
issues before continuing.
3. Start the dashboard
npm run ui
Open http://127.0.0.1:3030 in your browser. No login screen — the dashboard loads immediately.
4. First run in the dashboard
- Upload resume in the Get started section (step 1) — or fill profile in Settings
- Choose your goal — Find jobs, Upskill, or Change career (step 2)
- Click Try demo first to preview output without portals, or Start job search for a real run
- Enable Headed browser in Settings → Search & run when LinkedIn or Naukri login is required
See the full Dashboard guide for tabs, run history, resume/replay, and data paths.
5. Configure LLM (optional)
Add to .env:
OPENAI_API_KEY=sk-...
For Ollama on localhost:
OPENAI_BASE_URL=http://localhost:11434/v1
OPENAI_MODEL=llama3.1
Skip this step to use keyword-only mode — deterministic scoring with no API calls. See LLM providers.
6. Run your first pipeline
- Try demo first — sample jobs and shortlist without portal access
- Start job search (or upskilling / transition) — full pipeline from the Get started section
- Refresh the page during a run — progress stays connected; after a server restart, use Resume from checkpoint if shown
CLI alternatives
npm run search
npm run assist-apply
npm run demo
Windows PowerShell:
.\run-agent.ps1 -Mode search -Headed
macOS / Linux:
bash scripts/run-agent.sh both search
Outputs
Results are saved under data/output/:
jobs-{runId}.json,shortlist-{runId}.json,agent-report-{runId}.json- Tailored resumes in
data/output/tailored-resumes/ - Run logs in
data/ui/run-logs/; checkpoints indata/ui/checkpoints/ - CSV export from the Runs tab for any run with a shortlist
Docker (optional)
cp src/config.example.json src/config.json
docker compose up --build