compare

Open-source AI pentesting tools compared

Four questions separate these tools: what they need from you (source code or a URL), where you can run them, who pays for the model, and what the license lets you do.

Disclosure: this page is published by Xalgorix, one of the tools listed. Every fact comes from each project's own README or pricing page, checked on 19 October 2026, and each tool links to its source. Spotted something out of date? Tell us.

Summary

ToolLicenseGitHub starsApproachSelf-hosted needsHosted option
StrixApache-2.0~69.8kRuns your app dynamically and validates findings with proof-of-concept exploitsDocker plus an LLM API keyStrix Cloud: Pro $29 per seat per month; pentests billed per test
ShannonAGPL-3.0~49.1kWhite-box: reads the source code, then exploits the running app. “No exploit, no report”Your own model key, or local modelsKeygraph Pro: $50 per developer per month; Enterprise on request
PentestGPTMIT~16.0kAgentic penetration-testing framework powered by large language modelsYour own LLM providerNone listed
XalgorixApache-2.0~1.2kBlack-box: tests the running app; reports a finding only after exploiting it, then re-tests it with an independent verifierYour own LLM provider key, or local models via OllamaXalgorix Cloud: 10 credits for $10 (never expire); plans from $49/mo

Star counts from the GitHub API on 19 October 2026. For a closed-source reference point, XBOW does not publish prices; its pricing page describes usage-based pricing scoped to your environment. Xalgorix vs XBOW →

Which should you pick?

  • You have the source code and want CI gating on mergesShannon or Strix
  • You need to test what is deployed, including apps whose code you can't shareXalgorix
  • You want the largest community and the most integrationsStrix
  • You want to pay per scan, without seats or model keysXalgorix Cloud

Try Xalgorix on your staging app

10 credits for $10, no subscription, and credits never expire. Or self-host the open-source engine for free.