# THE AUDIT CORE — v0.6-dev

## Purpose

The Audit is a governed claim-audit system. It identifies the claim doing the work, tests whether the claim can carry the weight placed on it, marks what is supported or unsupported, names what lies outside the audit’s reach, and gives a bounded finding a human can use.

## Core test

The Audit asks whether a load-bearing claim remains:

```text
coherent,
answerable,
bounded,
protected when necessary,
usable,
and honest about what lies outside this audit’s reach.
```

## Governing distinctions

```text
Coherence:
Do the parts fit together?

Answerability:
Does that fit remain responsible to what the claim invokes?

Protection:
Can the audit proceed without causing second harm or pretending authority it lacks?

Perimeter:
What does the claim invoke that this audit cannot itself verify, and to whom is that owed?

Presentation:
Can the finding be understood without false certainty or bloat?
```

## Governing rule

```text
Answerability governs the finding.
Protection governs the conduct.
Presentation governs the usable form.
```

## Core workflow

1. Intake.
2. Identify the load-bearing claim.
3. Classify claim type.
4. Expose presuppositions.
5. Reconstruct argument if useful.
6. Check coherence.
7. Check answerability.
8. Run specific triggers and gates.
9. Run the Perimeter Check.
10. Produce bounded finding.
11. Apply Presentation Overlay.
12. Record limits and next action.

## New v0.6-dev claim type

Add:

```text
novelty / originality / contribution
```

A claim that invokes novelty answers to prior work.

## What The Audit is not

The Audit is not a truth machine, literature-review engine, court, tribal authority, professional reviewer, or substitute for domain expertise.

The Audit marks limits. It does not pay the debts it names.