Hand off work to AI agents in peace.

Durin governs AI agent access through MCP, with policies, human approvals, and audit records for requests routed through its gateway.

Book a demo Try without an account

Ship release 2.4

Writes allowed
Merges need approval

update_issue
create_pull_request
merge_pull_requestAllow once
send_message

Find gaps in your AI setup.

Answer ten questions or check a config. See your grade first.

Check my AI risk

Install Durin to your Agent

Copy a setup prompt into your Agent, or choose your client for manual setup.

Let your Agent set it up

Copy this prompt into your Agent. It runs the Durin CLI and guides setup. You choose your organization and Agent in the CLI and complete sign-in in the browser.

Set up Durin for my current Agent using pnpx @accessdurin/cli in an interactive terminal.

Guide me through login or signup and organization onboarding, then select my Agent in the CLI. Complete the MCP setup for that organization. Pause for my account choices, browser sign-in, any required payment setup, and MCP consent. Preserve my existing settings.

Run pnpx @accessdurin/cli verify, report the result, and tell me to reopen my Agent. If an interactive terminal is unavailable, give me the command to run myself.

Set up Durin for my current Agent using pnpx @accessdurin/cli in an interactive terminal.

Guide me through login or signup and organization onboarding, then select my Agent in the CLI. Complete the MCP setup for that organization. Pause for my account choices, browser sign-in, any required payment setup, and MCP consent. Preserve my existing settings.

Run pnpx @accessdurin/cli verify, report the result, and tell me to reopen my Agent. If an interactive terminal is unavailable, give me the command to run myself.

Manual setup

Choose your client and configure a direct MCP connection.

Use your personal endpoint

Copy your endpoint from Connected clients in Durin. Replace YOUR_PERSONAL_ENDPOINT_ID in the configuration with the value from your endpoint.

Get your MCP endpoint

Run in your terminal:

Open Claude Code and use /mcp to authenticate Durin.

Then request a tool through Durin and confirm the session in Connected clients.

Configuration for Claude Code

~/.claude.json

Merge the Durin entry into your existing configuration.

Give agents real work, not free rein.

Durin provides AI agent governance and MCP security. Set rules for connected tools, review sensitive actions, and follow each request through to its outcome.

Writes
Allow writes in one place
One decision covers every agent, on the connections you approve.
People review the changes that matter
Changes your rules send for review wait for a different person.
- update_issue
- merge_pull_request
- delete_channel
Every outcome is on record
Durin records who asked, which agent acted, the decision and the result.

Connections, not people.

First connection free. Your first four paid connections are $399 each per month. Each additional paid connection is $199 per month. 10 GB per connection monthly, pooled across your organization and prorated by enabled time. People and approved agent clients do not add seat charges.

See how pricing works

Before you hand off work.

Can agents make changes, not just read?

Yes. Requests that pass Durin’s checks go ahead by default, including changes. You decide in one place which changes need another person’s approval and which are denied.

What does Durin govern?

Durin governs requests sent through its gateway to configured connections. Direct API calls, browser actions, shell commands, and unmanaged agents that bypass the gateway need their own controls.

Can I try Durin before provider setup?

Yes. The free Durin sandbox lets you try policy, approvals, and audit before you connect company systems.

Hand off your first workflow.

Book a demo to explore agent access, policy checks, approvals, and audit records with our team.

Book a demo