Aegisora

Zero-trust proxy & escalation boundary for AI agents.

123 followers

Zero-trust proxy & escalation boundary for AI agents.

123 followers

Visit website

AI Infrastructure Tools

Stop selling abstract "AI safety". Enterprises buy operational control. Aegisora is an open-source, zero-latency proxy layer built for AppSec teams. Intercept malicious LLM actions, enforce least-privilege API access, mask PII on the fly, and generate readable audit logs for autonomous agents—without the bloated middleware.

This is the 2nd launch from Aegisora. View more

Aegisora Open Source

Launching today

Zero-trust proxy & escalation boundary for AI agents.

We are officially open-sourcing the Aegisora core engine! Stop breaking AI workflows with rigid firewalls. Aegisora acts as an escalation boundary, pausing risky LLM tool calls for human review with zero latency. Built for enterprise AppSec and Agentic AI builders.

Free

Launch tags:Open Source•Developer Tools•Artificial Intelligence

Launch Team / Built With

Aegisora

Maker

📌

Hey Hunters! 👋 Eray here, founder of Aegisora. After months of late-night debugging and building in public, I'm thrilled to officially open-source our core security engine today! 🚀 The Problem: > Traditional "allow/block" security gateways break non-deterministic AI agents. If an LLM makes a risky tool call and the firewall hard-blocks it, the entire execution loop usually crashes. The Solution: > I built Aegisora as a zero-trust proxy and an "Escalation Boundary." It dynamically intercepts risky LLM actions (like DB modifications), suspends the execution state, and routes it to a Human-in-the-Loop review dashboard—all without timing out the workflow. Since many of you are building the future of Agentic AI, I wanted to put this infrastructure in your hands. I'd love your honest feedback on our architecture and the GitHub repo. Feel free to ask me anything below, I'll be here all day answering your questions! 🛡️

Report

2d ago

Previous Aegisora Launches

AegisoraThe narrow control plane for AI agent tool and API calls.

Launched on August 5th, 2026

Forum Threads

p/aegisora

•

12d ago

How do you debug an autonomous AI agent that went completely off-script in production?

We ve all seen the dazzling demos of multi-agent systems executing complex workflows. But as developers start deploying these agents into deep production with direct API access, a terrifying reality sets in: debugging non-deterministic behavior.

When an agent hallucinates a tool call parameter, enters an infinite reasoning loop, or gets subtly manipulated by an indirect prompt injection, tracing why it happened at runtime is a nightmare. Traditional logs just show a wall of async LLM requests.

While building Aegisora, we realized that observability and security for agents can't be treated like traditional web apps you need real-time runtime inspection and deterministic guardrails.

I m curious to hear from builders here: What is your current strategy for monitoring and debugging autonomous agents when they misbehave in production? How do you catch runtime anomalies before they hit your external APIs?

Let's discuss in the comments!

p/aegisora

•

14d ago

What’s your worst "AI Agent gone rogue" moment in production?

As we move rapidly from experimental chat interfaces to fully autonomous agents with direct API access, runtime security is becoming a massive blind spot.

While building Aegisora (our open-source, zero-latency runtime proxy), I constantly hear terrifying stories from engineering teams about agents trying to execute unauthorized API calls, looping endlessly, or silently leaking PII because static guardrails and prompts failed.

View all

We ve all seen the dazzling demos of multi-agent systems executing complex workflows. But as developers start deploying these agents into deep production with direct API access, a terrifying reality sets in: debugging non-deterministic behavior.

When an agent hallucinates a tool call parameter, enters an infinite reasoning loop, or gets subtly manipulated by an indirect prompt injection, tracing why it happened at runtime is a nightmare. Traditional logs just show a wall of async LLM requests.

While building Aegisora, we realized that observability and security for agents can't be treated like traditional web apps you need real-time runtime inspection and deterministic guardrails.

I m curious to hear from builders here: What is your current strategy for monitoring and debugging autonomous agents when they misbehave in production? How do you catch runtime anomalies before they hit your external APIs?

Let's discuss in the comments!