Documentation

Documentation that explains the system the way the product does.

Learn Open Intent, integrate the authority runtime, verify tokens, and prove No Token, No Action with operator-grade documentation.

Start with the model Quickstart

Intent verificationAuthority TokensFail-closed enforcementEvidence export

70Docs entries

8Browse tracks

No TokenCore invariant

Browse the docs

Grouped by what you are trying to do.

Every page describes behaviour that runs in the platform today.

Getting Started

Understand what Intended is and submit your first intent for verification.

Architecture OverviewSystem components, data flow, service boundaries, and deployment topology for Intended, the authority layer fo↗What Is IntendedAn introduction to Intended, the authority layer for autonomous systems — verifying intent and issuing cryptog↗QuickstartSubmit your first intent to the Intended authority runtime, read the APPROVED / ESCALATED / DENIED outcome, an↗

Concepts

Understand the authority runtime pipeline, decision tokens, trust boundaries, and semantic intelligence.

Authority Runtime PipelineThe end-to-end path a request takes through Intended — interpret, evaluate, issue, audit — ending in a decisio↗Authority Token ModelThe Authority Token's real structure, Ed25519 signing, per-tenant keys, single-use nonce, 300-second lifetime,↗Policy Safety Controls — Structural Policy, Catastrophic Floor, Read/Write SignalHow Intended's authority engine decides on the raw structure of an action, enforces an always-on catastrophic ↗Semantic IntelligenceHow the Large Intent Model interprets an action into ranked hypotheses and risk signals — and how those signal↗Tenant Trust BoundaryHow Intended enforces tenant isolation through cryptographic trust boundaries and fail-closed defaults.↗Trust ModelUnderstand the trust model that underpins Intended, the authority layer for autonomous systems.↗

Operator Workflows

Author policy, simulate impact, deploy safely, and investigate incidents.

Author a PolicyAuthor an Intended authority policy as a governed draft — structure a policy set, scope its rules and effects,↗Control Center OperationsMonitor the authority runtime from the console control center and the supporting health, metrics, and live-dec↗Deploy and Rollback PoliciesPromote a reviewed policy draft to an active version and roll back to a prior version — the real /policy/deplo↗Incident ResponseContain and investigate an authority incident — query the SHA-256 audit chain, inspect and verify an Ed25519 d↗Simulate Policy ImpactPreview the effect of a policy change before deploying — read the draft impact summary (semantic diff, changed↗

Enterprise Administration

Manage tenants, roles, tokens, audit evidence, and emergency controls.

Audit ExportBuild, list, and download tenant audit evidence bundles — how the artifact is assembled, exactly how it is sig↗Emergency ControlsTenant containment actions — halt/resume a tenant or environment, revoke a credential — with exact action name↗Identity Provider OperationsInspect tenant identity readiness, run the OIDC/SAML SSO flows, and operate SCIM v2 provisioning — exact endpo↗Roles and Token GovernanceThe enterprise role/permission matrix and the API credential lifecycle — create, list, and revoke tenant API k↗Tenant AdministrationTenant-scoped admin endpoints for trust posture, environments, members, and role assignment — exact routes, re↗

Developer Integration

Integrate the authority runtime and verify tokens at your own boundary.

MCP Server and MCP GatewayThe Intended MCP server gives an AI assistant tools for asking Intended for decisions and, when you sign in as↗Agent Intent APISubmit an agent's proposed action to POST /v1/agent-intent for a governed decision. The response is a decision↗API QuickstartObtain an intended_live_ API key, submit your first intent to POST /intent, and handle the APPROVED (200), ESC↗Enforcement SDKThe @intended-inc/sdk reference — createIntendedSdk, submitIntent, verifyAuthorityToken, simulateIntent, evide↗Error Patterns and RemediationThe real error codes the Intended runtime returns — auth, validation, decision, token, and rate-limit failures↗LLM Gateway IntegrationGovern every agent tool call by pointing your runtime's model base URL at the Intended LLM gateway. One env va↗NVIDIA OpenShell and NemoClaw IntegrationCompile Open Intent intents and Domain LIM results into NVIDIA OpenShell sandbox policy YAML with @intended-in↗Runtime IntegrationsChoose how to put Intended in the path of an agent runtime — submit intents directly, verify the token at your↗Verify Decision TokensVerify Ed25519 Authority Tokens locally with @intended/verify — pin the kid, validate claims, enforce the 300s↗WebhooksGet a signed HTTP POST when a decision is made, escalated, approved, rejected, canceled or expires, or when an↗What your agent gets backALLOW, DENY and ESCALATE, and how an escalated action reaches its final answer. The SDKs, the MCP server, the ↗

Security & Compliance

Prove enforcement lineage, understand fail-closed controls, and run readiness checks.

Data Handling and RetentionWhat Intended stores for each decision, what it stores only as a digest, what it never stores, how long it kee↗Enforcement LineageHow Intended makes every authority decision provable — the per-tenant SHA-256 audit hash chain, chain verifica↗Fail-Closed ControlsHow Intended behaves when a decision cannot be established — default-DENY, the 500 AUTHORITY_LOOP_FAILED fail-↗How Intended Uses AIIntended is deterministic by default. No AI model decides, scores or influences an authority decision unless y↗NVIDIA OpenShell Integration Threat ModelThreat model and mitigation controls for Intended-governed NVIDIA OpenShell and NemoClaw runtime integrations.↗Operational ReadinessVerify Intended is healthy and its audit integrity holds before going live — liveness/readiness probes, a toke↗Third-Party Runtime BoundariesShared-responsibility and security boundaries for Intended integrations with NVIDIA OpenShell, NemoClaw, codin↗

Physical AI

Gate real-world actuation: classify goals, mint edge-verifier tokens, and run the ROS2/PLC integrations.

Physical Authority API ReferenceEvery endpoint, request field, response field, and failure mode for /v1/physical/* — ground truth for direct H↗What is Physical AI on IntendedHow the Intended Authority Runtime layers on top of an existing safety stack to classify, gate, and audit AI-i↗C++ Industrial ControlsIntegrate Intended into TwinCAT 3, Studio 5000, TIA Portal, or custom industrial PCs via the cloud HTTP API an↗Physical-AI QuickstartFrom sign-up to your first verified Authority Token for a physical action — register an actor, classify a stru↗PLC Migration PatternsCommon safety-PLC interlocks (Siemens F-CPU, GuardLogix, TwinCAT Safety, Pilz) translated into Intended policy↗Python ML PipelinesGate perception → policy pipelines (PyTorch / JAX) on Intended Authority Tokens using the Python SDK or the ra↗ROS2 IntegrationLayer Intended Authority Tokens onto a ROS2 agent — cobot, AMR, drone, surgical — using the intended-ros2 pack↗Safety-Case WritingAuthor a deployment safety case for a machine that adopts Intended. What Intended provides as mechanism, what ↗

API & CLI Reference

REST endpoints with real request/response shapes, and the flat CLI command reference.

API AuthenticationHow Intended authenticates API requests — credential types, the fail-closed gate, headers, scopes, rate limits↗API Reference: AuditQuery the per-tenant SHA-256 audit chain, verify its integrity, fetch offline-verifiable decision receipts tha↗API Reference: IntentsSubmit an IntentRequest to the authority runtime and get an APPROVED, ESCALATED, or DENIED decision with a sin↗Policies APIThe authority policy surface — evaluate an intent against policy with POST /authority/evaluate, manage policy ↗NVIDIA OpenShell CommandsCompile NVIDIA OpenShell runtime policy YAML from Open Intent intents and Domain LIM evaluations with @intende↗Policy CommandsInstall and inspect policy packs from the command line with @intended/cli — the real flat commands deploy-poli↗

Guides

Step-by-step guides for common platform tasks.

Deploy a PolicyInstall a policy pack for your tenant with the CLI or the API so your authority decisions reflect your rules —↗NVIDIA OpenShell Operations RunbookOperate a customer-run NVIDIA OpenShell / NemoClaw runtime governed by Intended — compile and apply policy, mo↗NVIDIA OpenShell Runtime QuickstartCompile a sandbox policy YAML for an NVIDIA OpenShell / OpenClaw / NemoClaw runtime from Intended presets, rev↗NVIDIA OpenShell Starter KitThe lightweight in-repo pattern for a first NVIDIA OpenShell pilot — the Open Intent taxonomy, the Intended CL↗Verify a TokenVerify an Authority Token before any downstream system acts on it — check the Ed25519 signature, the kid-pinne↗

Reference

Glossary, troubleshooting, persona workflows, and the go-live runbook.

GlossaryPrecise definitions of Intended terms — IntentRequest, authority decision, Authority Token, audit hash chain, ↗Go-Live RunbookThe production rollout sequence for Intended — pre-deployment verification, required controls, staged cutover,↗TroubleshootingSymptoms, causes, and fixes for common Intended issues — authentication errors, unexpected denials, token reje↗Platform Jobs-to-be-DonePersona-based guide to the Intended documentation and the workflows each role should follow.↗