Cencurity
Real-time security for AI-generated code inside VS Code.

The problem
AI coding tools generate code instantly.
But security checks happen too late — during review or after execution.
This creates a blind spot where insecure code can slip through unnoticed.
What Cencurity does
Cencurity sits between your IDE and the model.
It inspects generated code in real-time and blocks unsafe patterns before they reach your system.
What it does
- Opens the Cencurity Security Center inside VS Code.
- Routes supported LLM traffic through a local protection proxy.
- Keeps your existing provider API key where it already lives.
- Verifies routing automatically when protection is enabled.
- Shows live audit activity for real protected requests.
Quickstart
- Install the extension.
- Open Command Palette Ctrl+⇧Shift+P or ⌘Command+⇧Shift+P (macOS)
Cencurity: Enable Protection.
- Select your LLM provider.
- Open Command Palette Ctrl+⇧Shift+P or ⌘Command+⇧Shift+P (macOS)
Cencurity: Open Security Center.
That's it — protection is now active.
Features
Real-time Log Analysis

- Inspect generated code as it flows through the system
- See exactly what was detected and why
Dry Run Mode

- Simulate execution without risk
- Understand behavior before anything runs
Zero-click Attack Detection

- Detect dangerous patterns instantly
- Block risky operations like
subprocess, shell execution, etc.
Command Palette
Search for cencurity in the VS Code Command Palette to access the main actions:
Cencurity: Open Security Center — open the Security Center dashboard inside VS Code
Cencurity: Enable Protection — turn protection on and select your LLM provider
Cencurity: Disable Protection — turn protection off and restore previous supported routing settings
Cencurity: Test Protection — verify that requests are reaching the local proxy
Cencurity: Show Runtime Info — inspect the local runtime and protection state
Cencurity: Install or Update Core — install or refresh the local core runtime
Supported providers
- OpenAI
- Anthropic
- Gemini
- OpenRouter
- Other OpenAI-compatible LLMs
How it works
IDE → Cencurity Proxy → LLM Provider
- Your API key stays in your IDE.
- Requests are routed through a local proxy.
- Code is analyzed in real-time before execution.
What is CAST?
CAST (Code-Aware Security Transformation) protects a moment that existing tools don't cover.
| Model |
When it runs |
Main job |
Typical result |
| CAST |
while the model is still writing code |
stop unsafe output before it reaches the developer |
allow, redact, block |
| SAST |
after code already exists |
scan code for vulnerabilities |
findings after generation |
| DAST |
against a running app |
test runtime behavior |
runtime issues after deployment or staging |
| IAST |
inside an instrumented app |
watch real execution paths |
internal runtime findings |
The point is not that CAST replaces SAST.
The point is that CAST protects a different moment: while code is being generated.
Cencurity is the first tool built on CAST.
Notes
- Routing applies to supported env-based routing paths.
- Some extensions may bypass VS Code environment settings and not route through the proxy.
- The connector can use a local workspace runtime during development or a released core artifact in distribution builds.