Quickstart

Install & initialize Atryum

Download Atryum and register a server in Atryum for testing.

Download Atryum

curl -fsSL https://github.com/validmind/atryum/raw/main/install_atryum.sh | bash

The install script downloads the latest Atryum release and selects the correct binary for your operating system and CPU architecture. Supported platforms are Linux and macOS, each on amd64 (x86_64) or arm64 (aarch64).

Microsoft Windows is not supported natively — use Windows Subsystem for Linux (WSL) as a workaround. Alternatively, clone the repository and run Atryum with Docker — refer to the repository README for Docker Compose setup.

Set up Atryum test server

1. Generate a minimal testing configuration with a simple calculator Model Context Protocol (MCP) server that does not require any external credentials:

   ./atryum setup demo

   Copy
2. Start the Atryum service and register the test calculator server in Atryum's local database:

   ./atryum run --init-servers

   Copy
3. In your browser, navigate to localhost:8080 to open the Atryum local web user interface.
   

   Here, you can view servers, manage tool invocation approvals, configure rules, and more.

4. Within Atryum, click Servers in the left sidebar. Confirm that your test calculator server was successfully registered:
   • Verify that the server's connection and auth both display as ready.
   • Verify that Atryum exposed the server at localhost:8080/mcp/calc.
5. Connect your preferred coding agent to Atryum:
   • Open your agent's MCP settings and add a standard MCP server with the calc server address:

     http://localhost:8080/mcp/calc

     Copy
   • The agent will think it is talking to a calculator MCP server, but its tool invocations now pass through Atryum first.
6. Trigger a test tool call from your agent. For example:

   Use the calculator tools and show me 2*2

   Copy
7. Within Atryum, click Invocations in the left sidebar to review tool calls. Confirm that the calculator invocation request is Pending Approval — human approval is required by default.
   • Under Approval Required, select Approve to let the tool call run.
   • Verify that the invocation's status is Succeeded and that the approval was decided by a human.

Set up calculation server rule

Rules are if/then policies that tell Atryum how to handle tool invocations. For example — if the server is calc, then approve the call automatically.

Let's add a rule to control future matching tool calls:

1. In Atryum, click Rules in the left sidebar.
2. Click New Rule.
3. Under Action, select Auto Approve.
4. Under Servers / Sources, select calc to apply the rule only to calls to the test calculator server.
5. Make sure that Enabled is checked, then click Create Rule.
6. Try a calculator prompt from your agent again. Atryum should apply the new rule instead of treating the tool call like a brand-new manual decision:
   • Within Atryum, click Invocations in the left sidebar and confirm that your new call Succeeded.
   • Verify that the approval was decided by a rule.

Integrate Atryum

After setting up your test calculator server, connect Atryum with ValidMind and the rest of your tools to streamline your agentic oversight in one platform.

With ValidMind

Sync your ValidMind organization's agent records to Atryum to map tool invocations to agent records, use agent-scoped rules, and power invocation summarization.

Connecting agent records from ValidMind also allows you to use Atryum to evaluate tool invocations against each agent's charter, a plain-language policy that describes what an agent can do, what it cannot do, and which actions require human approval.

With coding agents

Connect your coding agents to Atryum, allowing Atryum to review tool invocations before your agents run them. Hooks and extensions are available for Claude Code, Cursor, Amp, Pi, and Codex.

With MCP servers

Register upstream Model Context Protocol (MCP) servers so Atryum can proxy tool calls to them. Atryum evaluates each call against your rules, holds upstream credentials on your behalf, and forwards approved calls to the server.

Connect MCP servers to add other upstream tools — such as local subprocess servers, remote HTTP servers with a static bearer token, or OAuth-protected services.

With local LLM providers

Set up local large language model (LLM) providers in Atryum to power AI evaluation rules and invocation summarization with models and credentials your team controls. Configuring LLM providers also allows you to use credentials and endpoints controlled by your team to evaluate tool calls against each agent's charter, a plain-language policy that describes what an agent can do, what it cannot do, and which actions require human approval.

Local LLMs are useful for bring-your-own-key setups, on-prem or air-gapped deployments, and teams that want to register multiple models and pick the right one per rule or feature.