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

# Overview

Security configuration covers caller identity, tool permissions, and data handling. VeADK provides authentication middleware, Agent Identity integration, content moderation callbacks, and Trusted MCP connections. Select the capabilities required by your application

## Capabilities

| Capability | Purpose | Integration point |
| - | - | - |
| [Inbound authentication](/productions/veadk/preview/en/components/security/inbound) | Verify callers of the application | Application ingress or gateway |
| [Outbound authentication](/productions/veadk/preview/en/components/security/outbound) | Provide credentials for tools calling external services | Function tools or MCP toolsets |
| [Content safety](/productions/veadk/preview/en/components/security/content-safety) | Moderate text around model and tool calls | Agent callbacks |
| [Trusted MCP](/productions/veadk/preview/en/components/security/trusted-mcp) | Connect to MCP services with remote attestation | MCP connection settings |

## Inbound and outbound authentication

Inbound authentication establishes who calls the application; the application must still authorize access to sessions and resources. Outbound authentication supplies credentials for third-party calls; the service and credential scope determine permission. Setting `Runner.user_id` does not perform either task

## Content safety

Moderation depends on configured policies and service responses. The current guardrail can continue execution when requests time out or HTTP errors occur. It should not be the only mandatory blocking control; see [Content safety](/productions/veadk/preview/en/components/security/content-safety)

## Trusted execution

Trusted MCP protects tool communication configured to use that connection. Model endpoints, other tools, logs, and storage require separate controls; enabling Trusted MCP does not automatically protect the entire application

See the [Agent Identity documentation](https://www.volcengine.com/docs/86848/2080920?lang=zh) for activation and permissions. Volcengine examples use resources in the selected region. BytePlus deployments need their own identity configuration and credentials, not Volcengine user pools or callback addresses