Skip to content

Open Intent Layer

The open standard for what agent actions are intended to do

The Open Intent Layer provides a common language for defining intent across agents — digital, physical, or hybrid — and the connectors, gateways, and enterprise systems they reach. v2.0: 29 domains, 173 categories, spanning enterprise digital operations and embodied / physical processes. Open source. Apache 2.0 licensed.

This page is about the standard itself. Intended operationalizes that standard with interpretation, verification, authority, and evidence.

The Open Intent Layer is one part of our open source effort. See the full open source ecosystem →

Open standard, operational platform

The standard is open. Verification is the system built on top of it.

The Open Intent Layer is meant to be portable and shared. Intended makes that standard operational at enterprise scale.

What the Open Intent Layer is

  • An open standard for describing what agent actions are intended to do
  • A portable language across runtimes, connectors, gateways, and enterprise systems
  • Apache 2.0 licensed, vendor-neutral, and designed to spread

What the platform adds on top

  • Large Intent Model interpretation across domains and systems
  • Intent Objects, Enterprise Capability Engine verification, and Authority Tokens
  • Operational workflows for escalation, replay, evidence, and enterprise controls

Why a shared intent layer matters

Without a common language, intent stays ambiguous.

Without a shared intent layer

POST /api/v2/deployments
gh.actions.workflow_run.completed
k8s.deployment.apps/v1.update
terraform plan -out=prod.tfplan

Every system names actions differently. Intent is ambiguous, evidence is fragmented, and controls do not travel cleanly across systems.

With the Open Intent Layer

sdlc.deploy.production
sdlc.ci.workflow-complete
sdlc.deploy.container-update
infra.iac.apply

One shared intent language. Portable controls. Clearer verification. Consistent evidence across platforms, frameworks, and integrations.

How Intended uses the standard

Open intent language, enterprise verification operations.

Operational interpretation

The platform interprets raw actions with the Large Intent Model and structures them into Intent Objects that can be verified, replayed, and audited.

Authority and enforcement

Intended turns the standard into an operational system with Authority Tokens, fail-closed execution boundaries, and managed decision workflows.

Evidence and governance

Teams need audit replay, evidence exports, approvals, compliance mappings, and shared governance workflows that go beyond adopting a taxonomy alone.

The taxonomy

29 domains, 173 categories

14 digital-operations domains and 15 physical / embodied-process domains. Search or browse the full Open Intent Layer taxonomy.

29 of 29 domains173 of 173 categories · OIL v2.0

Get started

Clone the taxonomy repository

The Open Intent Layer is distributed as source from the public GitHub repository. Clone it, reference the oil-taxonomy.json file, and adopt the stable OIL- codes in your logging, monitoring, policy, and audit surfaces. Apache 2.0 — free for any use, commercial or otherwise.

git

git clone https://github.com/intended-so/open-intent-layer.git

Open source

Apache 2.0

The Open Intent Layer is free to use, modify, and distribute. Use it in commercial products. Contribute back to the community.

A shared language for what agent actions are intended to do.

Open to adopt. Designed to spread.

Open Intent Layer — The Open Standard for Agent Intent | Intended