# Modei > Trust infrastructure for AI agents, cryptographic identity (Passports), policy enforcement (Gates), and tamper-proof audit trails (Attestations). Built for both agent operators and the services agents access. Modei lets AI agents prove who they are, what they're allowed to do, and produce cryptographic receipts for every action. Passports are signed credentials carried by agents. Gates are checkpoints that verify passports and enforce constraints before granting access. Every decision is recorded as a hash-linked Attestation that both sides can independently verify. ## Core Concepts - [Passports](https://modei.ai/docs/concepts/passports): Signed JWT credentials issued to agents. Carry identity, permissions, and constraints (spend caps, rate limits, domain allowlists). Verified by Gates in under a millisecond. - [Gates](https://modei.ai/docs/concepts/gates): Policy enforcement checkpoints that verify Passports and return allow, request_hold, or block before a protected action executes. - [Attestations](https://modei.ai/docs/concepts/attestations): Hash-linked, tamper-proof record of every Gate decision. Sequenced per passport so gaps are detectable. - [Issuers](https://modei.ai/docs/concepts/issuers): Organizational signing identities that authorize passport issuance. Ed25519 keypair, never stored server-side. - [Enforcement Layer](https://modei.ai/docs/concepts/enforcement-layer): Evaluates passport constraints (CEL). Called by resource servers on the backend, not by agents directly. - [Permission Catalogs](https://modei.ai/docs/concepts/permission-catalogs): Signed, versioned snapshots of a Gate's permissions and pricing. Immutable once published, so terms can't change without agent awareness. - [Trust Profiles](https://modei.ai/docs/concepts/trust-profiles): Security levels applied to a Gate, L1 Baseline (dev/low-risk), L2 Trusted (production), L3 Verified (sensitive operations, PoP required). - [Guardrails](https://modei.ai/docs/concepts/guardrails): Runtime constraint configuration for passports. - [Anonymous Access](https://modei.ai/docs/concepts/anonymous-access): How unauthenticated agents can access certain Gate operations. - [Supravision](https://modei.ai/docs/concepts/supravision): Human-in-the-loop approvals when a decision requires operator confirmation (SMS-based, no app required). ## Decision Values - allow: Action allowed. A signed attestation is issued. - request_hold: Action requires human approval. Do not proceed without operator confirmation. - block: Action violates a hard constraint. Do not proceed. ## Documentation - [Documentation home](https://modei.ai/docs): Overview, architecture, and where to start. - [Guide for AI Agents](https://modei.ai/docs/for-agents): Documentation written specifically for AI agents consuming Modei. - [Guide for Humans](https://modei.ai/docs/for-humans): No-code setup and concepts for human operators. - [Quickstart](https://modei.ai/docs/quickstart): 10-15 minute developer quickstart. - [Desktop Guide](https://modei.ai/docs/guides/desktop): Configure Modei for Claude Desktop and other MCP clients. - [Agent Autonomy Guide](https://modei.ai/docs/guides/agent-autonomy): Design patterns for autonomous agents. - [Passport-Only Guide](https://modei.ai/docs/guides/passport-only): Integration pattern using only passports (no gates). - [Templates](https://modei.ai/docs/templates): Pre-built constraint profiles and permission templates. ## API Reference Base URL: `https://modei.ai/api/v1` Authentication: `Authorization: Bearer mod_live_your_key_here` on all authenticated requests. Machine-readable spec: [OpenAPI 3.1](https://modei.ai/openapi.yaml) - [REST API Overview](https://modei.ai/docs/api): Endpoint index and conventions. - [Passports API](https://modei.ai/docs/api/passports): Issue, get, verify, revoke passports. - [Gates API](https://modei.ai/docs/api/gates): Create, configure, and test gates. - [Issuers API](https://modei.ai/docs/api/issuers): Manage organizational signing identities. - [Audit API](https://modei.ai/docs/api/audit): List and export attestations. - [Commerce API](https://modei.ai/docs/api/commerce): Catalogs, discovery, metering, settlement. ### Key Endpoints | Method | Path | Description | |--------|------|-------------| | GET | /discover | List discoverable Gates (no auth) | | GET | /gates/:id/catalog | Signed Permission Catalog for a Gate | | POST | /gates/:id/check | Verify a passport against a gate (resource-server backend) | | POST | /enforce | Passport-only enforcement (resource-server backend) | | POST | /passports | Issue a Passport | | GET | /passports/:id | Get a Passport | | GET | /passports/:id/verify | Verify a Passport | | POST | /gates | Create a Gate | | GET | /attestations | List attestations for your passport | | GET | /attestations/export | Export audit trail | ## SDKs & Integrations - [MCP Integration](https://modei.ai/docs/mcp): `npx modei-mcp`, 48 tools via Model Context Protocol. Works with Claude Desktop, Claude Code, and any MCP-compatible host. - [Python SDK](https://modei.ai/docs/sdk/python): Official Python client. - [TypeScript SDK](https://modei.ai/docs/sdk/typescript): Official TypeScript/JavaScript client. ## Personas - [Full Platform](https://modei.ai/docs/personas/full): Using all of Modei (issue + host). - [Issuer](https://modei.ai/docs/personas/issuer): Organizations issuing passports to their agents. - [Operator](https://modei.ai/docs/personas/operator): Services hosting gates and enforcing policy. ## Commerce - [Commerce Overview](https://modei.ai/docs/commerce): Agent-to-agent commerce infrastructure. - [Catalogs](https://modei.ai/docs/commerce/catalogs): Build and publish permission catalogs. - [Discovery](https://modei.ai/docs/commerce/discovery): How agents discover services. - [Billing](https://modei.ai/docs/commerce/billing): Consumption billing and settlement. - [Anti-Bait-and-Switch](https://modei.ai/docs/commerce/anti-bait-and-switch): Price change protections. ## Product Pages - [Passports](https://modei.ai/passports): Agent passports and portable credentials. - [Gates](https://modei.ai/gates): AI agent gates and access control. - [Enforcement](https://modei.ai/enforcement): Origin-side trust enforcement (CEL). - [Supravision](https://modei.ai/supravision): Human-in-the-loop oversight and approvals. - [Audit](https://modei.ai/audit): Tamper-proof audit trails and compliance logging. - [Commerce](https://modei.ai/commerce): Agent-to-agent commerce infrastructure. ## Pricing - [Pricing](https://modei.ai/pricing): Free tier (3 gates, 5 passports, 1k decisions/mo). Pro $29/mo (20 gates, 100 passports, 50k decisions). Teams $179/mo (unlimited gates & passports, 500k decisions). Enterprise custom. All plans include permanent, cryptographically signed transaction receipts. ## FAQ Q: What is Modei? A: Trust infrastructure for AI agents, cryptographic identity, policy enforcement, and tamper-proof audit trails so agents can act autonomously with verifiable credentials. Q: How do I integrate? A: Two primary paths. (1) MCP: install `modei-mcp` via npx for instant integration with Claude and other MCP hosts (48 tools). (2) REST API: call `https://modei.ai/api/v1` directly with a Bearer token. The Python SDK (`modei-python`) wraps the REST API; the TypeScript SDK (`modei-typescript`) is crypto-only (self-issued passports, signing, verification), call the REST API from TS with raw fetch(). Q: What's the difference between a Passport and a Gate? A: A Passport is a signed credential carried by an agent (identity + permissions + constraints). A Gate is a checkpoint deployed on a service that verifies passports and enforces constraints before granting access. Q: How are attestations tamper-proof? A: Every attestation is cryptographically signed and hash-linked to the previous one in the same passport's sequence. Both sides (agent and gate) hold a copy. Gaps are detectable; tampering is evident. Q: How much does it cost? A: Free to start (no credit card). Paid plans at $29 and $179/month. See /pricing for details. Q: Is Modei open source? A: Modei is a proprietary platform built on an open protocol for AI agent governance. The protocol defines the standard formats for passports, attestations, and permission catalogs. Q: Does verification require a network round-trip? A: No. Passport verification is local and cryptographic, sub-millisecond. Only decisions that touch shared state (spend caps, rate limits) hit the network. Q: How does Modei handle human approval? A: When a gate decision requires human approval (request_hold), Modei sends an SMS to the designated operator. Reply YES or NO directly, no app required. Emergency stop is available from the dashboard. ## Copy as Markdown Every documentation page on /docs supports copy-as-markdown for use with AI assistants. Look for the Copy button on doc pages. ## Optional - [Full text dump](https://modei.ai/llms-full.txt): Concatenated full text of all documentation, for agents that want everything in one request. - [Sitemap](https://modei.ai/sitemap.xml): Full site URL index. - [Blog](https://modei.ai/blog): Modei Journal, insights on AI agent trust, governance, and verification. - [About](https://modei.ai/about): About Modei. - [Contact](https://modei.ai/contact): Contact and support. - [Standard Logic Co.](https://standardlogic.ai): Parent company. - [Terms of Service](https://modei.ai/terms): Terms of Service. - [Privacy Policy](https://modei.ai/privacy): Privacy Policy.