Public Evidence Index

Mnemosyne is a proposed Layer-0 deterministic governance protocol downstream of generative systems. It does not generate content; it describes ways to evaluate outputs against versioned rules and produce reviewable evidence records.

This page does not claim production enforcement.

Legend

PUBLISHED EVIDENCE: an external record/file is linked.
PUBLICLY DEMONSTRATED: a public interface shows a bounded example, not live telemetry.
RESEARCH / ILLUSTRATIVE: a concept or scenario, not deployed behavior.

Protocol Explainer

GOVERNANCE METHOD

Describes a proposed downstream governance pattern for evaluating generated outputs.

Boundary: RESEARCH / ILLUSTRATIVE; no production enforcement claim.

DETERMINISTIC CONTRACT

Describes intended deterministic rule execution and bounded decision states in documentation.

Boundary: PUBLICLY DEMONSTRATED documentation framing; not a live deployment claim.

POLICY-AS-CODE

Describes versioned policy artifacts and change tracking as a method.

Boundary: RESEARCH / ILLUSTRATIVE unless independently evidenced per deployment.

FAIL-CLOSED GATE

Describes a fail-closed evaluation model for accepted/rejected outcomes.

Boundary: PUBLICLY DEMONSTRATED examples are bounded demonstrations, not live telemetry.

CANONICALIZATION / HASHING

Describes canonical serialization and hashing approaches for reviewable records.

Boundary: RESEARCH / ILLUSTRATIVE description; no guarantee of implementation scope.

ATTESTATION / RECEIPT

Describes evidence-style output records that can be reviewed as documentation artifacts.

Boundary: PUBLICLY DEMONSTRATED record style; no production guarantee or compliance claim.

CURRENT PUBLISHED PROTOCOL SPECIFICATION

Mnemosyne v3: Layer-0 Deterministic Governance Protocol for Generative AI

Version 3.0.0 · Date 2026-03-06 · Author Salman, Mert Kerem

A published protocol specification describing a proposed deterministic governance approach for generative-system outputs.

DOI: Open DOI link for v3 published record
Canonical record: Open canonical Zenodo record 18884426
File listed on record: WHITEPAPER.pdf

Verification steps: open the canonical record, confirm title/version/date/author, and inspect the listed WHITEPAPER.pdf.

Record and file publication are verified. Claims inside the paper are source-authored and are not independently validated by this website.

Last verified 2026-08-25.

EARLIER PUBLISHED APPLICATION PAPER

Mnemosyne Application Layer (v2.0): Deterministic IP Fidelity for Gaming & Ad-Tech

Version 2.0.0 · Date 2026-03-05 · Author Salman, Mert Kerem

An earlier published application paper focused on deterministic IP fidelity for gaming and ad-tech contexts.

DOI: Open DOI link for v2 published record
Canonical record: Open canonical Zenodo record 18869318
Public supplement: Open v2.0.0 public supplement release
Files listed on record: Mnemosyne-GameForge-v2.0.0.pdf and Mnemosyne-v2.0-GameForge.zip

Verification steps: open the canonical record, confirm title/version/date/author, and inspect the listed files.

Record and file publication are verified. Claims inside the paper are source-authored and are not independently validated by this website.

Last verified 2026-08-25.

DO NOT REPRESENT AS CURRENT IMPLEMENTATION.

Founder Binding

Mert Kerem Salman — Founder, Mnemosyne Labs; author name as published on the linked Zenodo records.

LinkedIn profile for founder contact context · Email founder directly

Operator documentation checked 2026-08-25

identities and behavior are volatile references, not a contract.

SEARCH

robots.txt is crawl guidance, not security or a license. Crawling does not guarantee indexing/citation/training.

ANSWER/CITATION

Public pages and linked records are the intended citation surface. Citation behavior depends on each external system.

USER-DIRECTED RETRIEVAL

Users may open the evidence links directly for manual verification of source records and listed files.

TRAINING

Crawling does not guarantee indexing/citation/training. Training decisions are controlled by external operators.

RESEARCH

Research / illustrative content is descriptive and not a deployment contract.

CONTROL/VERIFICATION

Technical posture: User-agent: * Allow: / with sitemap.

HUMAN ACTION READY via evidence links and pilot/contact.

AGENT ACTION PROTOCOL NOT OFFERED.