PROOF.md

ĀML Proof

Status: SHIPPED

If you only open one AML link, open this:

https://aruintelligence.github.io/aml-core/proof.html

ĀML is an interface firewall between AI/app intent and pixels.

The prototype rule used by this proof is:

render_allowed = restoration_value >= attention_cost

Reproduce a SUPPRESS decision

https://aruintelligence.github.io/aml-core/proof.html?attention=5&restoration=1

Expected: SUPPRESS.

Reproduce an ALLOW decision

https://aruintelligence.github.io/aml-core/proof.html?attention=5&restoration=5

Expected: ALLOW.

Prove replay locally

git clone https://github.com/aruintelligence/aml-core.git
cd aml-core
npm install --ignore-scripts
node demos/undeniable-proof/replay-proof.mjs
node scripts/check-flood-fixtures.js

CI requires deterministic replay and a balanced five-ALLOW / five-SUPPRESS fixture set.

Inspect more

• View Meaning: https://aruintelligence.github.io/aml-core/view-meaning.html
• Embed proof: docs/EMBED_PROOF.md
• Witness kit: docs/WITNESS_KIT.md
• Machine-readable proof: docs/proof-manifest.json

The proof demonstrates deterministic software policy behavior for declared inputs. It does not claim that attention_cost or restoration_value objectively measure human cognition or wellbeing.

Ask: change one value, copy the exact proof URL, screenshot the result, and file what happened.