> For the complete documentation index, see [llms.txt](https://docs.lpp-minduniverse.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lpp-minduniverse.org/lingua-pactum-protocol-lpp-documentation/4.-governance-objects/4.2-human-authorized-intent.md).

# 4.2 Human-Authorized Intent

The **Human-Authorized Intent** represents the human or institutional meaning from which action legitimacy derives.

It answers:

> **What did the legitimate authority actually authorize?**

This object exists because authority and meaning solve different governance problems.

Authority answers:

> **Who may legitimately authorize?**

Intent answers:

> **What was actually authorized?**

A system may possess a valid Authority Object while producing an action that no longer reflects the purpose, target, limits, or material meaning originally authorized.

Therefore:

```
Valid Authority
≠
Preserved Authorized Intent
```

#### Why Intent Must Be Explicit

Consider a valid institutional authority authorizing:

```
Isolate compromised Host A.
```

An agent trajectory may later evolve into:

```
Isolate Host A
        ↓
Modify Host A firewall
        ↓
Modify subnet policy
        ↓
Change organization-wide network rules
```

The original authority source may still be valid.

But the final action may no longer preserve the authorized meaning.

The question is therefore not merely:

```
Does valid authority exist?
```

It is also:

```
Does the current action still mean
what the legitimate authority authorized?
```

#### Human Does Not Mean Only an Individual Person

The term **Human-Authorized Intent** includes legitimate intent established through recognized human-governed institutions.

The origin may therefore be:

* an individual,
* an authorized organizational role,
* an institution,
* a regulated entity,
* or another recognized legal authority.

The key requirement is that the authorized meaning remains grounded in human-recognizable governance rather than becoming self-generated by the AI system.

#### Intent Is Not a Prompt

A prompt may contain authorized intent.

But the two concepts are not equivalent.

A prompt may also contain:

* irrelevant context,
* formatting requests,
* examples,
* system instructions,
* model-generated text,
* retrieved data,
* or later semantic contamination.

Human-Authorized Intent therefore represents the **governance-relevant meaning**, not simply the raw input string.

```
Raw Prompt
≠
Human-Authorized Intent
```

#### Intent Is Not Execution Intent

Human-Authorized Intent exists upstream.

Execution Intent represents the concrete action proposed later.

The relationship is:

```
Human-Authorized Intent
        ↓
task evolution
        ↓
Action Candidate
        ↓
Execution Intent
```

The later action must remain reconcilable with the upstream authorized meaning.

#### Current Public Specification Boundary

GitBook v2.0 treats Human-Authorized Intent as a canonical governance object.

Its public relationship to Authority Object, Semantic Verification, and re-admission is defined here and in Chapter 6.

The detailed machine-readable public schema is reserved for the Governance Object Schemas section of the Appendix.

This documentation does not infer additional normative fields beyond the current public specification.