We find the problems and we fix them.

What Each Agent Could Do

Netlumi shows what each agent can read, change or delete.

Netlumi tells you which data an agent could change that you could not get back, and what to do first.

AI Attack Paths

Each path starts where untrusted text can reach an agent, such as a public endpoint, a document or a message. It runs through the agent and its tools to your data.

Netlumi shows the change that closes each path, starting with the one that closes the most. Where it can, the fix arrives as a pull request.

Runtime Gate

Business and Enterprise plans · opt-in

An optional gate can stop an agent action that would destroy data you cannot restore. Netlumi never reads the prompt or the conversation.

You choose which agents it covers, and you can run it in record-only mode first. Every allow and block is shown with its reason.

Agent Activity

AWS · Growth, Business and Enterprise plans

Netlumi follows what each agent's identities do in your AWS accounts. It does not read the agents' conversations.

The Activity page shows the week's alerts and gate decisions. An alert is raised only for high-severity events, such as a blocked action.

Decoys

Growth, Business and Enterprise plans

Decoys tell you within minutes if an attacker finds them or an agent is being steered by something it read.

Each alert names the address the use came from.

Model and Endpoint Security

Bedrock and SageMaker are cloud resources and can be misconfigured like any other. Azure OpenAI and Google Vertex AI are detected in code.

We check every model and endpoint you run for public exposure, roles that let anyone invoke them, invocation logging that is switched off, and unencrypted training data.

Shadow AI

Model calls and API keys are often added to services without a security review.

We read your repositories and find AI usage that is not in your AI inventory, with the file and line where it appears.

Infrastructure as Code

The free scan finds misconfigurations in your IaC code, Kubernetes manifests and Dockerfiles. In connected repositories, Netlumi checks Dockerfiles today and reads your IaC code to trace cloud findings to their file; infrastructure-as-code checks are coming.

Secrets

AWS access keys, private keys and API keys for AI model providers committed to a connected repository. The free scan also finds many other common key and token formats. Each finding says what to do about it. Secrets in git history are coming.

Dependency Vulnerabilities

The free scan finds known CVEs in lockfiles and manifests for JavaScript, TypeScript, Python, Go, Java, Rust, Ruby, PHP and .NET, and your dependency list is not sent to a third party. In connected repositories, Netlumi checks the base image of each Dockerfile for known high and critical vulnerabilities; dependency checks are coming.

Code Checks

The free scan finds security issues in application code for Python, Go, JavaScript and TypeScript. Findings in example, test and fixture code are listed separately and not counted. These checks are coming to connected repositories.

Security Detections

Continuous checks against good practice and CIS Benchmarks™ recommendations, at the same depth in every account you connect. Findings are mapped to compliance frameworks (see Compliance).

Attack Path Analysis

Netlumi maps the chain from an exposure, through identity escalation, to your data, so you can start with the paths an attacker could use rather than ordering work by severity label alone.

Drift Detection

Netlumi shows when live infrastructure stops matching its IaC code, what changed, who changed it, and whether the change went through your pipeline.

Code-to-Cloud Mapping

Netlumi traces every resource defined in your IaC code back to the file, line and commit that created it, so you know where a fix goes and which team owns it.

Live Changes and Threats

Netlumi's rules run on the cloud events your accounts forward, alongside GuardDuty, the AWS detector you already use. Results are shown in one place, with a live change log of who did what, where and when.

Only high-severity threats raise an alert. Everything else is shown in your risk view without notifying anyone.

Context and Audit Record

An alert on a resource shows the attack paths that run through it, so you can see at once whether a suspicious change opened a route to your data.

Every alert is resolved with an outcome and notes. The alert register records when each one happened, was detected, notified, acknowledged and resolved, and downloads as CSV for your auditors.

Posture by Framework

See where you stand on SOC 2, ISO 27001, PCI DSS, HIPAA, NIS2 and the rest, which controls are failing, and which findings and resources are behind each one.

Security and Compliance Report

A report with a compliance section, plus PDF and CSV evidence of the fixes you merged: what was wrong, the pull request that fixed it, and when. Netlumi Lens answers questions about your framework mappings in plain English.

Fixes as Pull Requests

Netlumi writes the fix and opens a pull request in your repository. Your engineers review it like any other change.

• Review mode by default: nothing is written to your repository until you approve the fix.
• Fix pull requests are checked before you see them, and never applied for you.
• Each pull request says what was checked and what was not.
• Fixes that would destroy or replace a resource are blocked.

Netlumi Lens

Ask questions in plain English about your findings, attack paths and framework mappings, such as who can read the customer bucket, what changed in the last day, or which paths reach production.

Answers cite the records they come from, so you can check them. You do not need a query language or a spreadsheet export.