What Strix does
Strix (strix.ai) runs autonomous AI agents that attack your own web apps, APIs, repositories and cloud setup the way a pentester would, then keep only the findings they can reproduce with a proof-of-exploit. The vendor lists coverage for REST, GraphQL and gRPC APIs, pull request reviews in CI, cloud and Kubernetes misconfigurations, and an auto-fix step that writes a patch, retests it and opens a pull request. The open-source project's README lists test categories including broken access control, SQL/NoSQL/command injection, SSRF, XXE, RCE, XSS, CSRF, business logic flaws, authentication issues and API security.
There are two ways to use it. The open-source CLI (usestrix/strix, Apache-2.0) installs with one curl command, needs Docker running and an API key for an LLM provider (the README names OpenAI, Anthropic, Google, Vercel AI Gateway, OpenRouter, Vertex AI, Bedrock, Azure and local models via Ollama or LM Studio), and runs with strix --target <path or URL>; headless mode exits non-zero when it finds vulnerabilities, so it can gate a GitHub Actions pipeline. The hosted platform at app.strix.ai adds scheduled pentests, attack surface monitoring, one-click autofix and Jira, Linear and Slack integrations, and its Enterprise tier can run in your VPC, on-premise or air-gapped. Strix says source code is never stored or used for model training and lists SOC 2 Type II and ISO 27001. The pricing page footer names OmniSecure, Inc.
