Products

Theme

Stop AI Agents fromRunning Wild

Give your agents identity and limits. Shield your tools and services from unauthorized access. Every decision signed and verifiable.

Get Started Free See How It Works

No credit card required.

Built for Both Sides of the Agent Stack

Whether You Deploy Agents or Host What They Access

Agent Operators

If You Deploy AI Agents

Your agents represent you. Give them cryptographic identity so every service they touch knows exactly who sent them and what they're authorized to do.

Issue passports with scoped permissions and spend limits
Define guardrails that travel with the agent everywhere
Monitor every action through a real-time audit trail
Revoke access instantly if something goes wrong

Service Providers

If You Run APIs, MCP Servers, or Tools

Stop trusting agent requests blindly. Require cryptographic proof of identity and authorization before granting access to anything.

Verify agent identity at the edge on every request
Publish permission catalogs so agents know what’s available
Enforce rate limits, spend caps, and action boundaries
Maintain a tamper-proof record of every interaction

How It Works

From Zero to Verified in Four Steps

Add trust infrastructure to your agent stack in minutes, not months.

Register Your Identity

Create your organization on Modus and generate cryptographic credentials for your agents or your services.

Issue Passports or Create Gates

Deploying agents? Issue passports with scoped permissions. Hosting services? Create gates that define what’s allowed.

Connect via MCP or API

Plug in with the TypeScript or Python SDK. Modus is MCP-native — it works with Claude, GPT, and any MCP-compatible host.

Every Action Signed & Verified

From this point forward, every agent request is verified, every decision is signed, and every action is recorded.

gate.ts — Protect a resource in 12 lines

{
  "mcpServers": {
    "modus": {
      "command": "npx",
      "args": ["modus-mcp"],
      "env": {
        "MODUS_API_KEY": "mod_live_xxxxxxxx"
      }
    }
  }
}

22 MCP tools·modus-mcp on npm·REST API

What's Inside

Five Pillars of Trust

Passports

Cryptographic identity bound to what the agent actually is — model, version, architecture. Not just who operates it. Issue in seconds. Verify in under a millisecond. Delegate with guaranteed attenuation.

Learn more about Passports

Gates

Policy enforcement where it matters — between the agent and the resource. Permission catalogs, spend limits, rate controls, domain allowlists. Credentials pinned to specific catalog versions so terms can’t change without your awareness.

Learn more about Gates

Enforcement

Decisions made before damage is done. Spend limits enforced at the gate, not tracked after. Revocation propagated in real time, even mid-execution. Four-decision model: permit, block, suspend, or output-hold.

Learn more about Enforcement

Supravision

Human oversight from any device. See what your agents are doing in real time. Approve or deny suspended actions from your phone. Panic-stop a rogue agent on a remote server. Guardian mode for child safety. Managerial mode for operational oversight.

Learn more about Supravision

Audit

Every action produces a signed receipt. Every receipt is hash-linked to the last. Both sides hold a copy. Gaps are detectable. Tampering is evident. Export the whole chain for compliance, accounting, or regulatory review.

Learn more about Audit

Built for Production

Everything you need to secure agent actions at scale.

Prove Which Agent Made the Request

Every passport, attestation, and catalog is cryptographically signed with Ed25519. Verify trust without a network call.

Set Rules for What Agents Can Do

Spending caps, rate limits, domain allowlists, approval workflows. 17 built-in constraint evaluators, all enforced automatically.

e.g., cap spending at $50/day, only allow calls to Stripe and GitHub

Choose the Right Security Level

L1 for development, L2 for production with trusted issuers, L3 for high assurance with proof-of-possession.

e.g., require proof-of-possession for production APIs

Keep Signed Audit Logs

Every enforcement decision creates a signed attestation. Sequential numbering prevents gap-filling. Export anytime.

Works with Claude, GPT, and Any MCP Client

First-class Model Context Protocol support. 22 tools available out of the box.

Purpose-Built Protocol

A governance protocol designed specifically for autonomous AI agents. Not a wrapper around existing standards — a new foundation for execution-grade identity.

Command Center