v3.9.1
VulnRadar documentation
Paste a URL, get a ranked list of what is wrong with it and how to fix each one. These pages cover the REST API, webhooks, quotas, self-hosting, and the internals if you want to add a check of your own.
- checks
- 905+
- categories
- 18
- ports swept
- 145
- API version
- v3
First scan
Scan one URL, no account
Open /demo and run a scan. It is the same engine as everything below; an account adds your own history, an API key, and higher limits.
Create an account
Sign up on the hosted instance, or self-host and register the first user through the normal signup form.
Generate an API key
Profile, then API Keys, then Generate New Key. The raw key is shown once and never again. How many active keys you may hold is set by your plan: 1 on Free, 3 on Core Supporter, 10 on Pro Supporter, unlimited on Elite Supporter.
Send the first scan
POST the target to /api/v3/scan with your key as a Bearer token. A bare hostname works: https:// is prepended for you.
Read the findings
Each finding carries a stable id, a severity, the evidence that triggered it, and fix steps with a copyable snippet.
curl -X POST "https://vulnradar.dev/api/v3/scan" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "example.com", "portScan": true}'portScan is optional. Leave it out and only the web checks run. Full request and response shapes are on the API reference.
The documentation set
Start here
- Introduction
- What the scanner does and how to get a first result
- Setup Guide
- Install, migrate the database, run it locally
- Browser Extension
- Scan from the toolbar, auto-scan, on-page alerts
Using VulnRadar
- Scheduled Scans
- Recurring scans on a schedule, with regression alerts
- Triage & Remediation
- Track findings from open to fixed, plus support tickets
- Teams
- Roles, invitations, and sharing scans across a team
- Sharing & Public Pages
- Share links, the public directory, host reports, badges
- Account Security
- 2FA, sessions, social logins, and your data
- Plans and Billing
- Plan limits, credits, upgrading, and cancelling
- Troubleshooting Scans
- Failures, timeouts, blocked targets, and empty results
Integrations & AI
- GitHub Scanning
- Scan a connected repo for secrets and code flaws
- AI Features
- Vera chat, AI verification, summaries, and BYOK keys
Running your own
- Self-Hosting
- docker-compose, TLS, backups, upgrades
- Administration
- The admin panel, staff roles, backups, and retention
- Configuration
- How configuration resolves, and the values you will change
API reference
- API Reference
- Endpoints, request and response shapes, error codes
- API Playground
- Send live calls and copy them as code in your language
- Reports & Compliance
- Export SARIF, PDF, Markdown, or a compliance crosswalk
- CLI
- Run scans from the terminal and gate CI on findings
- Webhooks
- Discord, Slack, and plain JSON delivery
- Rate Limits
- Daily quotas, per-IP limits, and the 429 contract
Internals
- Architecture
- Request lifecycle, subsystems, and the check registry
- Developer Documentation
- Local workflow, scripts, and building an SDK
What gets checked
905+ detections live in lib/scanner/checks-data/, one JSON file per category, each paired with a detector module in lib/scanner/checks/. Every check has a stable id, so a finding you triage today keeps the same id on the next scan and in the API response.
The 18 categories are active-probes, api, client-side, code, configuration, content, cookies, dns, email, headers, host-validation, information-disclosure, reputation, secrets-extended, ssl, supply-chain, tls, vibe-code. Pass scanners on a scan request to run a subset.
The port sweep is separate and opt-in behind the single portScan boolean. It opens a bounded TCP socket against each of 145 curated well-known ports, reads whatever greeting comes back, and reports reachability and version disclosure per service. It does not depend on the URL scheme: SSH on port 22 is probed the same way whether you asked for an https:// target or not.
Beyond the check catalogue, every scan also fingerprints the software the host runs (server, framework, CDN, analytics, and client-side libraries) and correlates any version it can read against known CVEs through OSV.dev and the NVD, enriched with CISA KEV and FIRST.org EPSS. A vulnerable version raises one aggregated finding that lists its CVE IDs.
The full catalogue is served, unauthenticated, from GET /api/v3/finding-types. Use it if you are building an SDK and want every id ahead of time. See Developer documentation for the payload shape.
Keeping pages out of a scan
When VulnRadar crawls a site for a multi-page scan, its crawler identifies itself as VulnRadar and reads /robots.txt before discovering pages. To keep specific paths out of a VulnRadar scan, add a group that names VulnRadar with Disallow rules:
User-agent: VulnRadar
Disallow: /checks
Disallow: /generated/Only a group that names VulnRadar specifically is honored. A blanket User-agent: * rule does not fence the scanner out, so a site’s general bot policy never quietly narrows a security scan you asked for. Rules are matched as standard robots.txt path prefixes.
This affects page discovery only. Search engines follow their own * rules, so anything you disallow for VulnRadar stays fully indexable for them. And a URL you enter directly is always scanned, robots.txt or not: the rule shapes what the crawler wanders into, not what you deliberately point it at.
Support and versions
If something here is wrong or missing, say so. Bug reports and doc corrections go to the issue tracker; anything account-specific goes through the contact form. Legal terms, the privacy policy, and the acceptable-use rules for scanning targets you do not own are on the legal pages.
- app
- 3.9.1
- engine
- 3.3.2
- api
- v3