v2.0 — Open Source & Free

Every AI agent has a brain.
We make sure nobody poisons it.

The security layer for any AI agent's memory. Like Cloudflare, but for everything your AI remembers — regardless of platform.

npx shieldcortex setup
View on GitHub

How It Works

A defence pipeline that sits between your AI agent and its memory. Every write is scanned. Every read is filtered. Everything is logged.

Agent → ShieldCortex → Any Memory Backend
↓
Scan → Score → Classify → Audit

Defence Layers

Memory Firewall

Catches prompt injection, hidden instructions, encoding tricks, and command injection.

Free

Audit Logger

Full forensic trail of every memory operation. What was stored, when, and why.

Free

Trust Scorer

Scores memories by source reliability. User input scores highest, agent-generated lowest.

Free

Sensitivity Classifier

Detects passwords, API keys, PII — auto-redacts on recall.

Pro

Fragmentation Detector

Catches multi-step assembly attacks spread across days of fragmented memories.

Pro

Attack Vectors Blocked

Direct Injection

Hidden [SYSTEM] instructions disguised as normal content

Credential Harvesting

Memories that try to exfiltrate API keys or passwords

Encoding Tricks

Base64, hex, and unicode payloads that bypass text filters

Slow-Burn Assembly

Attack fragments planted over days that combine into exploits

Privilege Escalation

Memories referencing system commands, file paths, or admin URLs

Get Protected in 30 Seconds

Install ShieldCortex and scan your existing memories for threats.

$ npm install -g shieldcortex
$ npx shieldcortex setup
# Then ask your AI agent:
"Scan my memories for threats"

Trusted by Developers

Join the growing community protecting their AI agents.

"Finally, someone is taking AI memory security seriously. ShieldCortex caught three prompt injections we didn't even know were there."

— AI Engineering Lead

"Setup took 30 seconds. The audit log alone is worth it — we can finally see what our agents are remembering."

— DevOps Engineer

"The trust scoring system is elegant. It fundamentally changes how we think about agent data provenance."

— Security Researcher

Works With Any AI Agent

ShieldCortex is agent-agnostic middleware. If your agent has memory, we protect it.

OpenClaw

Native hook support

Moltbot

Plugin integration

Claude Code

Hooks + MCP server

LangChain

Memory wrapper API

AutoGPT

Memory backend plugin

CrewAI

Agent memory layer

MCP Agents

Any MCP-compatible agent

Custom Agents

npm API / REST API

Protect Your AI Agent Today

Free and open source. MIT licence. No account required.

npm install -g shieldcortex