Govern every agent you run.

One install. Every agent governed. Works with any MCP-compatible environment, any language, any framework. Integration is not the same as governance: the install brings an agent into the layer, and policy decides every action it takes from then on.

Claude Code, Cursor, Codex, Pi, Hermes, OpenClaw, VS Code, or any MCP-compatible environment.

Zero Friction

Add one line. Done.

Works in any MCP-compatible environment. No terminal, no changes to the server you run: paste the config and your agent launches the governed path itself.

Step 1

Copy this configuration:

{
  "mcpServers": {
    "your-server-governed": {
      "command": "npx",
      "args": ["-y", "agentomy-mcp-gateway", "--", "<your-mcp-server-command>"],
      "env": {
        "AGENTOMY_ENDPOINT": "https://api.agentomy.com",
        "AGENTOMY_API_KEY": "<your-agentomy-api-key>"
      }
    }
  }
}

Step 2

Paste into your agent environment settings:

CursorClaude CodeCodexPiHermesOpenClawVS CodeOther

Paste into Cursor MCP settings at Settings > MCP Servers

Step 3

Run your agent. It is governed.

From paste to governed: 90 seconds.

Full Control

For developers who want full control.

TypeScript or Python. Import, configure, govern.

TypeScript

// npm install agentomy-agent
import { GovernancePipeline } from 'agentomy-agent'

const pipeline = new GovernancePipeline({
  endpoint: process.env.AGENTOMY_ENDPOINT,
  token: process.env.AGENTOMY_TOKEN,
  agentId: 'my-agent'
})

await pipeline.evaluate({ action: 'data_export' })

Python

# Python guardrail: available in a hosted workspace.
from agentomy_guardrail import AgentomyGuardrail

guard = AgentomyGuardrail(
    endpoint=os.environ['AGENTOMY_ENDPOINT'],
    agent_id='my-agent'
)

result = guard.check(action='data_export', tier='builder')

Benchmark First

Run the benchmark first. See the gap. Then install.

See exactly what governance your agents are missing. Then install Agentomy to close every gap.

# Run GovernanceBench against your agents
npx governancebench run --target <url>

# See your scores across 6 dimensions
# Then install to close the gaps
npm install agentomy-agent

Wrap what you run

Govern an MCP server you already run.

The gateway sits between your agent and any MCP server. Every tool call is checked against your policy before it runs, and a refusal is enforced, not just logged. The server is never changed.

Step 1

In your MCP config, launch the server through the gateway:

{
  "mcpServers": {
    "dev-tools": {
      "command": "agentomy-mcp-gateway",
      "args": ["--", "your-mcp-server", "--its-flags"],
      "env": {
        "AGENTOMY_ENDPOINT": "https://your-agentomy",
        "AGENTOMY_API_KEY": "..."
      }
    }
  }
}

Step 2

Or wrap it at the command line:

agentomy-mcp-gateway -- your-mcp-server --its-flags

Step 3

Run your agent. Every tool call is governed. The server never knew.

What happens next

Run your agent. Every action it takes will be logged and checked against your policies before it runs. You can ask AgentomyGuide anything about your agents at any time. No logs. No dashboard. Just ask.