Security for AI agents
Your agent remembers.
Make sure it remembers safely.
ShieldCortex guards the three surfaces every AI agent exposes — what it stores, what it does, and what it sees. Poisoned memory, hostile tool calls, and injected web content are stopped before they become your agent's truth.
MIT licensed · open source Reproduce locally: npx shieldcortex scan-skill examples/adversarial-skill/
The problem
Most memory tools help agents remember more.
None of that matters if what they remember is a lie.
The moment an agent gets persistent memory, tool access, and a web connection, it exposes three attack surfaces — and a single poisoned write, hijacked tool call, or injected page can quietly become permanent, trusted context.
Hostile content written into memory today is recalled as truth in every future session. The agent doesn't just make a mistake — it learns one.
Prompt injection can steer an agent into running commands it was never asked to run. Guardrails on what it remembers mean nothing without guardrails on what it does.
Every fetched web page and tool result is untrusted input. Hidden instructions in the outside world become inside instructions the moment the model reads them.
The product
Three surfaces, one shield
01 · Stores
Memory Firewall
Every memory write runs a six-layer defence pipeline — injection, hidden instructions, credential leaks, and poisoning are caught before they reach storage. Recall is inspectable: see what was stored, why it ranked, and what was refused.
Memory docs →02 · Does
Iron Dome
Gates every tool call at runtime. Catastrophic actions — disk wipes, curl | sh, secret exfiltration — are hard-blocked; risky ones route through operator approval. Benign work is never interrupted.
03 · Sees
Environment Firewall
Scrubs fetched pages and tool output before the model reads them, so prompt injection hidden in the outside world never silently becomes an instruction. X-Ray inspects packages and files for covert payloads.
See X-Ray →Proof, not promises
Watch it block a live attack
A prompt injection and privilege escalation, caught in real time. Thirty seconds, no signup.
Caught pre-CVE
Flagged axios 1.14.1, plain-crypto-js and color-diff-napi before they reached CVE databases.
27 injection patterns
Eight attack categories scanned on every prompt, tool call, and memory write — extensible with your own rules, free.
Every decision logged
A full audit trail of what was stored, recalled, blocked, and why — exportable, reviewable, replayable session by session.
Quickstart
Protected in two commands
OpenClaw
Restart the gateway and every session is protected.
Node.js / CLI
Wires into Claude Code, Codex, and MCP agents. Dashboard included.
Python
LangChain and CrewAI extras. Python docs →
Node.js 18+ or Python 3.9+ · macOS, Linux & Windows · free and open source
Start protecting your agents
Full protection is free and open source. Enterprise adds fleet management, memory sync at scale, and SSO for teams and servers.
Running a team or a fleet? Talk to us about Enterprise