> ## Documentation Index
> Fetch the complete documentation index at: https://docs.integratewise.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# 16 · Governance & Authority View

> May it be done? Identity → Authority → Capability → Policy → Plan → Approval → Authorization → Execution → Evidence — before Hermes ever executes.

## May It Be Done?

Governance answers **one question**: *may it be done?*

```text theme={null}
Identity
   ↓
Authority             (who/what may act in this Work context)
   ↓
Capability            (what is being requested — provider-neutral)
   ↓
Policy                (what constraints apply)
   ↓
Plan                  (capability sequence + inputs + scope)
   ↓
Approval              (human or delegated authorization)
   ↓
Authorization         (governance verdict: allow / deny / require approval)
   ↓
Execution             (Hermes — only on allow)
   ↓
Evidence              (what happened, durably)
```

<Callout type="warning">
  Governance verdicts are **durable, auditable, and traceable** to identity, policy, and Work. Hermes never executes without one.
</Callout>

## What Governance Evaluates

* **Who** is requesting — user, Twin, or agent identity, with what roles and delegation.
* **What** is being requested — which capability, with what inputs and side-effects.
* **In what Work and Business Context** — not ambiently; every plan is scoped to a Work and a Business Context slice.
* **Under which policy** — tenant, Work-type, and capability policies, including approval thresholds.
* **With what evidence shape** — what receipt is expected, and how verification will judge outcome.

## Verdicts

| Verdict              | Meaning                                                            |
| -------------------- | ------------------------------------------------------------------ |
| **Allow**            | Plan is authorized; Hermes may execute.                            |
| **Deny**             | Plan is not authorized in this context; no execution.              |
| **Require approval** | A human (or delegated approver) must approve before authorization. |

Approval is itself a Work action with its own identity, evidence, and audit trail.

## Separation That Matters

```text theme={null}
Capability Fabric          Governance                Hermes
 "What can be done?"  →  "May it be done?"  →  "Execute it."
       │                       │                      │
   ADK / MCP               Policy /               Provider
  resolution               Approval               execution
```

Hermes does **not** become an authority of its own. Replacing Hermes does not change what is allowed — only how allowed work is executed.

## Where to Go Next

<CardGroup cols={2}>
  <Card title="Hermes Execution" icon="bolt" href="/platform/hermes-execution-view">
    How authorized plans are executed with evidence.
  </Card>

  <Card title="Identity & Authority" icon="key" href="/platform/identity-authority-view">
    The identities and tenancy governance reasons about.
  </Card>

  <Card title="Verification & Reconciliation" icon="check-double" href="/closure/verification-reconciliation">
    How outcomes are judged and what becomes Spine truth.
  </Card>
</CardGroup>