This repository contains two independent public-reference full reports for Neotro Protocol, an observer-layer protocol designed to organize state-transition signals between existing AI systems and human-review processes without replacing, modifying, or automatically controlling the original system.
The reports demonstrate how Neotro Protocol can be applied to public datasets in generative AI safety taxonomy and AI NPC dialogue domains while preserving the boundary between observation and operational decision-making.
· reports/Neotro_Aegis_2_0_Safety_Taxonomy_Observer_Mapping_Full_Report_v1_1.pdf
Generative AI safety-taxonomy observer mapping using Aegis 2.0.
· reports/Neotro_LIGHT_AI_NPC_State_Transition_Observation_Full_Report_v1_4.pdf
AI NPC / roleplay / world-grounded dialogue state-transition observation using LIGHT.
· PUBLIC_POC_REPORT_SERIES_OVERVIEW.md
Overview of the public PoC report series, experiment positions, and reserved follow-up layer.
· PUBLIC_BOUNDARY_NOTICE.md
Public-release boundary and out-of-scope notice.
· RELEASE_NOTES_v1_0.md
Release notes for this public PoC report series.
· CITATION.cff
Citation metadata for the report series.
· LICENSE.md
Neotro Protocol Public Reference License Notice v1.0.
· CHECKSUMS_SHA256.txt
SHA256 checksums for release files.
· Experiment 1 — LIGHT
AI NPC / game-dialogue state-transition observation.
· Experiment 2 — Aegis 2.0
Generative AI safety-taxonomy observer mapping.
· Experiment 3 — Reserved follow-up layer
ToxicChat / RealToxicityPrompts or partner-specific datasets may be used later as a separate stress-boundary and human-review escalation observation layer.
Neotro Protocol complements existing safety classifiers, moderation systems, game systems, policy engines, and human reviewers by adding an observer-layer review reference over existing data.
All report outputs should be read as public-reference observer-layer review states. Certified classifier performance, moderation benchmark results, production enforcement claims, source-system replacement, and operational control remain outside this public-reference scope.
These reports use public-reference datasets and aggregate analysis outputs. Long raw text excerpts, deployment-specific materials, partner-calibrated materials, implementation-sensitive observer materials, operational-use implementation materials, and production-use materials remain outside this public release.
GitHub Release:
https://github.com/Neotro-Engine/neotro-ai-domain-public-poc-report-series/releases/tag/v1.0
Zenodo DOI:
https://doi.org/10.5281/zenodo.20100495
Research / Business Inquiry:
neotroprotocol@gmail.com