Quick Start

Quick Start

Get your first scan running in under 5 minutes.

1. Initialize a repository

sentinel init /path/to/your/repo

This creates:

• sentinel.toml — configuration file with documented defaults
• .sentinel/ directory — database, reports, and output
• Adds .sentinel/ to .gitignore

Choose a profile for quick setup:

sentinel init /path/to/repo --profile minimal   # no LLM needed, heuristic-only
sentinel init /path/to/repo --profile standard   # all detectors, basic LLM
sentinel init /path/to/repo --profile full       # all detectors, enhanced analysis

2. Run your first scan

sentinel scan /path/to/your/repo

Output:

Scan complete: 12 findings in run #1
  Severity: 3 medium, 9 low
Report: /path/to/your/repo/.sentinel/report-1.md

No LLM? Add --skip-judge --skip-llm to run with only deterministic detectors. No model provider needed.

Using a cloud provider instead of Ollama? Configure it in sentinel.toml — see Model Providers for OpenAI and Azure setup.

3. Review the report

Open the generated markdown report, or launch the web UI:

sentinel serve /path/to/your/repo

This opens a browser-based dashboard at http://127.0.0.1:8888.

4. Triage findings

# Suppress a false positive
sentinel suppress 3 --reason "Intentional complexity in parser"

# Approve a finding for GitHub issue creation
sentinel approve 7

5. Create GitHub issues (optional)

export SENTINEL_GITHUB_TOKEN=ghp_...
export SENTINEL_GITHUB_OWNER=your-username
export SENTINEL_GITHUB_REPO=your-repo

sentinel create-issues --dry-run     # preview first
sentinel create-issues               # create for real

What's next?

• Configuration — customize detectors, models, and thresholds
• Detectors — learn what each detector finds
• Scheduling — set up overnight scans with cron
• Morning Report — understand the report format