> ## Documentation Index
> Fetch the complete documentation index at: https://docs.goantiai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Blast Radius

> See exactly what an agent can reach if compromised.

The **Blast Radius** view answers: *"If this agent is compromised, what can an attacker reach?"*

It shows an interactive graph with the agent at the center, every tool and service the agent has access to, and color-coded risk levels:

* **Green** — read-only, low risk
* **Yellow** — write access, medium risk
* **Red** — admin or delete access, high risk

## How to use it

1. Open any agent → **Security** tab
2. Hover over any node to see scope details
3. Pan and zoom to explore complex permission graphs

## Reducing blast radius

A large blast radius means a compromised agent can do a lot of damage. Consider:

* Splitting the agent into multiple agents with narrower permissions
* Tightening the policy to remove unused scopes
* Setting a shorter custom token TTL for high-risk scopes