AWS Features Strix in Its GLM 5.3 Launch on Amazon Bedrock

AWS featured Strix in its official launch announcement for GLM 5.3 on Amazon Bedrock, showcasing our open-source AI pentesting agents in an end-to-end security testing walkthrough.

We believe cybersecurity is the most important use case for AI right now. But a capable model is only part of the equation. Turning that capability into useful security testing means giving agents the tools to investigate applications, attempt exploits, and verify what they find. That’s what we’re building with Strix.

From attack surface to working proof of concept

In the AWS walkthrough, Strix agents test a locally running application. They map the attack surface, investigate potential vulnerabilities, and attempt to validate findings with working proofs of concept. Reports include severity, supporting evidence, and remediation guidance.

Why we’re building in the open

We believe the tools you trust to test your systems should be open to inspection. You should be able to understand how the agents work, adapt them to your environment, and decide where they run.

That’s why Strix is open source. Security researchers and developers can inspect the code, contribute improvements, and build on the same foundation rather than starting from scratch.

Open-weight models like GLM 5.3 extend that choice to the model powering the agents. Run Strix on your own infrastructure with Bedrock handling inference, or connect it to a self-hosted model for a fully on-prem deployment. The latter lets you run both the security agents and model inference within your own environment.

Our thesis is simple: adopting more capable AI security should not mean giving up visibility into the tools, flexibility to adapt them, or control over where they run.

Inference through your AWS account

With Bedrock, you can use GLM 5.3 through managed APIs, with inference under your AWS account’s controls and no inference infrastructure to manage.

Strix uses your existing AWS credentials, including CLI profiles or IAM roles, so there’s no separate model-provider API key to manage.

Try it yourself

You’ll need Strix installed with Bedrock support and an AWS account with access to GLM 5.3. The AWS setup guide uses the explicit inference-profile configuration below. Replace YOUR_REGION and YOUR_ACCOUNT_ID with your AWS details.

Use your own application or a target you have permission to test.

Thanks to AWS for featuring Strix and Z.ai for building GLM 5.3.

Read the full AWS announcement for the setup guide and video walkthrough.