Pentes
Pentes.io is an attack surface monitoring platform. It runs non-destructive security scans against assets you have cryptographically proven you own, then uses AI to triage the findings into a clear, prioritized, interactive report — giving security teams continuous visibility into their own perimeter between annual penetration tests.
What is Pentes? Pentes.io is an attack surface monitoring platform that runs non-destructive vulnerability scans against assets a customer has cryptographically proven they own, then uses an LLM to triage the structured findings into a prioritized, interactive report. It is a monitoring and assessment product — not an exploitation tool — giving security teams continuous visibility into their own perimeter between annual penetration tests.
What Makes It Powerful
Ownership-Gated Scanning
Cryptographic proof of asset ownership is required before any scan. The control plane refuses to enqueue a job until the proof exists and is current, and the worker re-verifies ownership immediately before each scanner runs.
Non-Destructive by Architecture
Scans run with nuclei (intrusive and exploit templates excluded), OWASP ZAP in passive mode, and testssl.sh. No exploit payloads, no brute-force, no fuzzing — the dangerous templates are removed from the worker image, not just toggled off.
AI-Assisted Triage
Anthropic Claude reads structured SARIF findings, deduplicates across scanners, prioritizes by real-world impact, and writes plain-English fix guidance. The model sees findings only — never your secrets, credentials, or systems.
Continuous Scan Diffs
Track your perimeter over time. Pentes diffs every run to show new, persistent, and fixed findings, turning point-in-time assessments into continuous monitoring between annual penetration tests.
Interactive Reports & Audit Log
Get interactive HTML reports with deduplication and prioritization, plus JSON, Markdown, and SARIF exports for downstream tooling. Every verification, authorization, and scan event is recorded in an immutable, append-only audit log.
Glass-Box AI & Tenant Isolation
The triage LLM has no shell, no network access to targets, and no tool-use channel. Every repository method is scoped by organization and PostgreSQL row-level security enforces the same boundary at the database layer — cross-tenant reads are impossible at two independent layers.
Built For Your Industry
Security Teams
Maintain continuous visibility into your external perimeter between annual penetration tests. Catch newly exposed services, expired certificates, and misconfigurations as they appear — not months later in the next audit.
Developers & Vibe Coders
Ship fast without leaving exposure behind. Verify ownership of your domain, scan after every deploy, and get prioritized, plain-English fixes you can act on without a dedicated security engineer.
Agencies & Consultants
Monitor the assets you manage on behalf of clients. Owner-verified scans, immutable audit logs, and exportable reports give you defensible, repeatable evidence of the security posture you deliver.
Frequently Asked Questions
Is Pentes safe to run against production systems?
Yes. Pentes is non-destructive by architecture. Exploit, intrusive, DoS, and fuzzing templates are removed from the scanner container image itself — not just disabled by a runtime flag — so the constraint cannot be bypassed by configuration. It uses nuclei (with intrusive templates excluded), OWASP ZAP in passive mode, and testssl.sh for TLS posture.
How does Pentes prevent unauthorized scanning?
A scan never runs until you have proven ownership of the target via a DNS TXT record or HTTPS path challenge — the same pattern used by ACME and Let's Encrypt. The scan worker re-reads the ownership token from the database immediately before any scanner starts, so stale state in the queue cannot sneak a scan through.
What does the AI actually do?
Pentes uses Anthropic Claude to triage scanner findings. The model reads only SARIF JSON plus minimal asset context — never secrets, credentials, or system access. It deduplicates findings across scanners, prioritizes by real-world impact, and produces plain-English fix guidance. The LLM has no shell, no network access to targets, and no tool-use channel.
What do I get in a report?
Interactive HTML reports with cross-scanner deduplication, impact-based prioritization, and clear remediation guidance. Continuous scan diffs track new, persistent, and fixed findings between runs, and every verification, authorization, and scan event is written to an immutable, append-only audit log. Reports export to JSON, Markdown, and SARIF for downstream tooling.
How much does Pentes cost?
Pentes offers a Free tier with 5 scans per month, a Vibe Coder plan at $14.99/month for 100 scans, and a Pro plan at $99/month for 1,000 scans. Sign up with email and password or Google Sign-In.