Skip to content
View LortuArte's full-sized avatar

Block or report LortuArte

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
LortuArte/README.md

πŸ›‘οΈ AEGIS Core | Enterprise AI Agent Firewall

PyPI Version License: MIT

AEGIS Demo

(Click above to watch AEGIS block a 1,000-thread concurrent loop attack in <1ms)

The Sub-millisecond Tool-Execution Firewall for Enterprise AI Agents. Cloud HTTP Gateways (like LangSmith) protect your LLM tokens. AEGIS protects your Stripe account and Crypto Wallets from asynchronous Agent Double-Spending.


🚨 The 1,000-Thread Double Spend Vulnerability

When an autonomous agent enters a hyper-cognitive loop, it can fire hundreds of concurrent tool calls (e.g., Stripe API charges or Web3 transactions).

Cloud-based guardrails suffer from network latency (~50ms - 150ms). By the time the cloud gateway registers the budget depletion, the in-flight transactions have already drained your accounts.

⚑ Why AEGIS? (Zero-Latency Policy Enforcement)

AEGIS is a horizontal L3 Policy Gate designed purely for speed and state locking. It sits exactly between your Agent's reasoning engine and your execution tools.

  • Sub-Millisecond Concurrency: High-Frequency Local IPC Memory Locks (< 1ms) budget resolution.
  • Infrastructure Agnostic: Works with LangChain, AutoGen, CrewAI, or raw Python scripts.
  • Zero Dependencies: No Redis, no Kafka. Pure Python in-memory atomic locks.

πŸ“Š Benchmark Proof (Concurrent Drift Attack)

In a live stress test simulating a rogue agent attempting 1,000 concurrent Stripe tool calls:

  • βœ… 1000 concurrent requests intercepted
  • βœ… 0 double spends (Zero Financial Drift)
  • βœ… $50,000 overspend prevented
  • βœ… <0.001ms lock latency per request

πŸ“¦ Quickstart & Installation

pip install aegis-core-lortuarte-sdk

πŸ› οΈ Proof of Concept: The 3-Line Integration Wrap your high-risk tools (payments, trades, database writes) with the AEGIS gate.


# 1. Initialize local IPC Client
aegis_gate = PolicyGate(daily_budget_usd=100)

def execute_agent_payment(agent_id, amount):
    # 2. Intercept budget spending BEFORE tool execution (<1ms lock)
    decision = aegis_gate.evaluate_tool_execution(
        agent_id=agent_id,
        operation="stripe_charge",
        amount=amount
    )
    
    if decision["status"] == "ALLOW":
        # Safe to execute real API call
        # stripe.Charge.create(...)
        return "Transaction Authorized"
    else:
        # Loop blocked instantly. Budget saved.
        return f"BLOCKED: Asynchronous Double-Spend Prevented in {decision['latency']}ms"

🧠 The Architecture (vs. LLM Gateways)

Feature LangSmith / Portkey (Cloud) AEGIS Core (Local IPC)
Primary Target Token Spend / Prompt Injection Tool Execution / Money Spend
Latency 50ms - 200ms (HTTP) < 1ms (In-Memory)
Double-Spend Protection Fails under high concurrency Atomic deterministic locking

πŸš€ Enterprise Pilot

Running autonomous agents in production?

Without runtime security, a single race condition can cause overspending, double-spends, and unsafe execution.

AEGIS Enterprise includes:

  • πŸ“Š Policy dashboards
  • πŸ‘οΈ Real-time observability
  • 🀝 Enterprise SLA support
  • ⚑ Custom infrastructure integrations

Book Enterprise Pilot at: https://aegis-api.com/

Pinned Loading

  1. aegis-sdk aegis-sdk Public

    Sub-millisecond Tool-Execution Firewall for AI Agents. Prevents asynchronous double-spending via Local IPC Locks.

    Python 1

  2. aegis-web-api aegis-web-api Public

    HTML 1