Product

Safety outside the agent. Security in the action path.

IntentFrame sits between your AI agent and the systems it can change. It checks every proposed action against deterministic limits and plain-English policy before anything touches money, data, customers, files, infrastructure, or APIs.

Runtime authorization for AI agent actions.

What it is

IntentFrame is a runtime enforcement layer. It receives proposed actions from an agent, evaluates each action against your policy, and returns a clear allow or block decision before the action can affect a real system.

What it is not

IntentFrame does not make the agent smarter, write better responses, or guarantee the agent always makes the best business decision. It makes sure an unsafe or unauthorized action does not execute.

IntentFrame's job is to protect your business, not the agent.

We are a veto engine, not an optimization engine.

How it Works

The agent never grades its own homework.

The agent proposes what it wants to do and why. IntentFrame checks that proposal against your policy before the action runs. Sensitive actions are carried out through a governed path the agent does not control, so a tricked or confused agent cannot move money, change customer data, or mutate a system on its own say-so.

Agent proposes IntentFrame checks Allowed action runs Audit record

If the agent is tricked, confused, or malicious, the action still has to pass an external boundary first.

Intent vs. Permission

What are you trying to do, and why?

Traditional security asks
Is this action allowed?
  • Permissions check
  • Rule matching
  • Blind to context and meaning
IntentFrame asks
What are you trying to do, and why?
  • Intent validation
  • Policy against business meaning
  • Context-aware judgment

Every attempted action is evaluated for:

Intent alignment

Does this action actually match the request?

Scope boundaries

Is the agent operating where it is allowed?

Attack indicators

Does this resemble injection or manipulation?

Risk exposure

Are the consequences acceptable?

If validation fails, the action does not execute.

If anything is unclear, execution stops.

Nothing passes silently.

Security is the default state.

Prevention First

IntentFrame is structural prevention.

Agents cannot act directly on sensitive systems. Every action must pass validation before it touches your money, data, customers, files, or APIs.

Surveillance

Watch agents as they act. Detect problems. Alert. Respond.

The agent already has capability when you notice something wrong.

Structural prevention

Agents cannot act directly. Every action must pass validation before execution.

Unauthorized actions stop before they reach production systems.

Structural preventionSurveillance (monitoring)
Credential accessOnly the validated execution path has credentialsAgents have credentials
Attack timingPrevents execution capability from being misusedDetects attacks after capability exists
Defense typeArchitectural — novel attacks still hit the boundaryPattern-based — can miss novel attacks
Response modelProactive enforcement before executionReactive alerting
Security outcomeStops what should not happenLogs what happened

Monitoring tells you what happened.

IntentFrame controls what is allowed to happen.

Hard limits first. Meaning where it matters. Enforcement always.

1

Hard limits

Amount caps, allowed recipients and accounts, allowed action types, and obviously unsafe actions are checked instantly, with no AI involved.

2

Action check

IntentFrame looks at what the action would actually do, including whether the stated reason matches the real details of the request.

3

Plain-English policy

Rules that depend on meaning are checked against your written policy. Example: "Refund only genuine manufacturing defects under $100."

4

Enforced outcome

Allowed actions are carried out through the governed path. Blocked actions never run.

5

Decision record

Every allow or block result can be recorded with context, policy version, rationale, and timestamp.

Meaning is separated from authority.

Hard limits are checked first, with no AI. When a decision depends on meaning, IntentFrame separates understanding from authority: one step works out what the action really does, and a separate step decides it against your policy. The agent's words are treated as evidence, never as instructions.

Hard limits

Fixed rules like amounts, recipients, and allowed actions are enforced before any AI runs.

Analysis

Works out what the proposed action would actually do. No authority to approve.

Guardian

Decides against your policy. No reason to obey the agent's request.

Everything from the agent is treated as untrusted evidence, not authority.

A brilliant judge with no locked door is theater.

Judgment only matters if something enforces it. Sensitive actions run through a governed path that the agent cannot bypass, so a compromised agent still cannot directly spend money, leak data, or change production systems on its own.

  • Credentials kept out of the agent
  • No bypass path
  • Policy versioning
  • Audit trail

Scope

Our job is to protect the business.

IntentFrame does not try to make your agent more charming, more creative, or more profitable. It makes sure a bad, confused, or compromised agent cannot cross the boundaries you set.

Our job

  • Stop unauthorized refunds and payments
  • Block data leaks and system damage
  • Prevent unapproved promises
  • Keep sensitive actions inside the governed path

Not our job

  • Write better emails
  • Make the model smarter
  • Guarantee the most profitable choice
  • Replace your product or application logic

Making the agent good is the developer's job. Making sure a bad agent cannot hurt the business is ours.

Inspect the boundary before you trust it.

IntentFrame is developed through the IntentFrame GitHub organization. The core runtime, SDKs, policy model, Hermes Agent plugin, tests, and documentation are open for review.

  • 18 lockstep-versioned PyPI packages.
  • Hermes Agent plugin available.
  • Public documentation and examples.
  • Policy and audit architecture inspectable.

Neutral by Design

The control point is not locked to one model, framework, or cloud.

IntentFrame is designed to sit between agents and actions across stacks. Use the SDK, an HTTPS authorization endpoint, a tool gateway, or a custom adapter. The enforcement boundary stays independent of the agent framework.

  • Model-agnostic
  • Framework-agnostic
  • Cloud-agnostic
  • Works with SDK, HTTPS, MCP/tool gateways, and custom adapters

Use the SDK, call the API, or wire IntentFrame into your agent stack.

Python Actor SDK

For Python agents that can import IntentFrame directly.

HTTPS Authorization Endpoint

For non-Python stacks. Send proposed tool calls over HTTP and receive allow/block decisions.

Hermes Agent Plugin

Install IntentFrame as a security plugin for Nous Research's Hermes Agent, the self-improving AI agent. Govern terminal, code, file, patch, and scheduled actions before they run.

MCP and Tool Gateways

Apply IntentFrame between an agent and the tools it can call.

Custom Agents

Integrate with in-house frameworks by wrapping the action path.

Security Invariant

Fail-closed by default.

Any ambiguity results in a block or a route to human review — never silent approval. IntentFrame is designed to overblock when uncertain rather than let unauthorized actions through.

Any ambiguity results in rejection or escalation — never silent approval.

Validation unavailable? Execution halts.
Intent unclear? Execution halts.
Pattern unexpected? Execution halts.
System under uncertainty? Execution halts.

Security is the default state.

What IntentFrame protects, and what it does not.

Protects

  • Unauthorized actions before execution.
  • Agent attempts to act outside policy.
  • Credential exposure through direct agent access.
  • Business-rule violations that require semantic judgment.
  • Audit gaps around what an agent proposed and why it was allowed or blocked.

Does not protect

  • It does not make the base model smarter.
  • It does not guarantee the agent writes better messages.
  • It does not prevent the model from being tricked internally.
  • It does not replace application security, identity, or infrastructure controls.
  • It does not remove the need to define policy.

We do not claim to stop prompt injection. We stop the resulting action at the boundary.

Ready to put a boundary around your agent's actions?

Tell us what your agents can do today and what systems they can touch. We will help map the enforcement path.