> ## 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.

# Trusted MCP

Trusted MCP adds remote attestation and end-to-end encryption to a configured MCP connection. Use it with tool services that support the corresponding trusted protocol

## Scope of the trusted connection

VeADK connects to Trusted MCP services through `TrustedMcpToolset`. With confidential-computing environments such as Jeddak AICC, the client verifies the remote service against its attestation policy and establishes a trusted channel for that MCP connection. Configure protection separately for the agent runtime, model services, and other data destinations

## Prerequisites

<Steps>
  <Step title="Prepare a Trusted MCP service">
    Obtain a Trusted MCP service address, e.g. `https://your-trusted-mcp.example.com/mcp`.
  </Step>

  <Step title="Prepare the AICC config">
    Prepare the AICC config file `./aicc_config.json` (for the confidential environment and remote attestation); see the [official example](https://github.com/volcengine/AICC-Trusted-MCP/blob/main/README.md).
  </Step>
</Steps>

## Usage

Turn on `x-trusted-mcp: true` in the connection parameters, then connect with `TrustedMcpToolset`:

```python title="agent.py" lines theme={null}
import asyncio
import os
from google.adk.tools.mcp_tool.mcp_session_manager import StreamableHTTPConnectionParams
from veadk import Agent
from veadk.tools.mcp_tool.trusted_mcp_toolset import TrustedMcpToolset

async def main():
    toolset = TrustedMcpToolset(
        connection_params=StreamableHTTPConnectionParams(
            url=os.environ["TRUSTED_MCP_URL"],
            headers={"x-trusted-mcp": "true"},
        ),
    )
    try:
        tools = await toolset.get_tools()
        print("Available tools:", [tool.name for tool in tools])
        agent = Agent(name="trusted_assistant", tools=[toolset])
        print(await agent.run("Describe the tools available to you"))
    finally:
        await toolset.close()

asyncio.run(main())
```

## Options

Key `TrustedMcpToolset` options:

| Option | Location | Default | Description |
| - | - | - | - |
| `x-trusted-mcp` | Header | Unset | Whether to enable Trusted MCP and open the trusted channel |
| AICC configuration | File path | `./aicc_config.json` | AICC config file path, for attestation and policy in the confidential-computing environment |

For the config file's format and details, see the [Trusted MCP config reference](https://github.com/volcengine/AICC-Trusted-MCP/blob/main/README.md).

## Connection requirements and verification

Configure the model, set `TRUSTED_MCP_URL` to a Trusted MCP Streamable HTTP endpoint, and prepare the AICC configuration required by your provider. Running `python agent.py` should list available tools and then print a model response

The trusted channel requires both `StreamableHTTPConnectionParams` and the `x-trusted-mcp` header with the string value `true`. Omitting the header or using SSE/stdio parameters does not enable this channel. AICC settings belong to the trusted client configuration, not a `TrustedMcpToolset(aicc_config=...)` constructor argument. Attestation policies must match the service requirements

Successful tool discovery only verifies connectivity and discovery; also check the trusted client's attestation result. This channel does not protect data in other tools, model services, or logs