Strix 1.0.4 setup: Docker and API model requirements

Strix is an open source AI penetration testing tool that the speaker describes as using multiple agents to find and exploit application vulnerabilities. Agents handle reconnaissance, write exploits and validate results in parallel. The speaker says Strix produces working proof-of-concept code, suggests remediation patches and can run in CI/CD on pull requests. These are presented as capabilities, but the transcript does not document a specific successful exploit against the speaker's app.

The installation walkthrough reports Strix version 1.0.4 after a pip command. Its command line accepts a URL, GitHub repository, local folder, domain or IP address, with quick, standard and deep scan modes. Each AI agent runs in an isolated Docker sandbox. The described toolkit includes an HTTP intercepting proxy, a browser, a shell and an exploit runtime. Targets include broken authentication, IDOR, business logic flaws and API security issues.

Running the framework on your own machine does not make its model inference local. The speaker explicitly says the intelligence comes from an external model API and that deep scans consume paid tokens. Docker must also be running before a scan can start. The video closes with a restriction: test only applications you own or have explicit written permission to test.