The full instance001 repository directory, published on the FMI site.

An open architecture for host-side memory, relational reasoning, earned abstraction, assumption tracking, and cognitive depth control. AGPLV3

Open repository

Start Here.

Open repository

Fractal Media Infrastructure is an independent public-interest organization for open AI research, local-first tooling, and public education. Home of the instance001 R&D lab and the Let's Rethink AI media branch.

Open repository

Local first, cloud optional LoRA trainer for Wan 2.1 models currently, focused on AMD Windows. Tested on RX 9060 XT (8GB half). Supports image/video pipelines; more models/audio planned. Sister to Chatty-art. End to end achievable fully local. Cloud when you need it, local when you don't. AGPLv3.

Open repository

Local first, cloud optional image, GIF, video and audio generator. Drop in a GGUF, type one sentence, get media. No API keys required (cloud when you need it, local when you don't), no node graphs. Vulkan-ready. Plain English UI anyone can use. Built on llama.cpp + stable-diffusion.cpp.

Open repository

A barebones, inspectable Rust specimen of Entropy Folding: vault failures, triangulate recurring blockers, metabolise them into reusable constraints, and reduce future search chaos.

Open repository

The everyday use variant of the chatty ai exoskeleton series. Chatty-cog (short for cognitive prosthetic) features sandbox, local networking, peer to peer functions, standard handshake plug for module expansion allowing for functionality expansion. GPU main Ai, CPU only bookkeeper, tri-helix memory system allowing for cross module context awareness

Open repository

Local-first Android AI chat for small phones, with local GGUF and optional BYO cloud models, dual AI roles, persistent context, and a user-owned sandbox.

Open repository

A skills-based framework for working effectively with any AI tool. Teaches the underlying mechanics of human-AI interaction - not a prompt list, not copy-paste templates, but transferable, epistemic principles that work across models, tools, and tasks.

Open repository

Model-native persistent memory patterns for LLMs: semantic buckets, reducer-governed state, hot context injection, and context routing packets for compact continuity without cold-log token waste.

Open repository

A local-first nanochat fork with a guided builder dashboard, GGUF helper assistant, and ChattyCog module packaging aka The best ChatGPT that $0 and a lot of patience can buy.

Open repository

A modernized, provider-neutral prompt and interaction scaffolds for shaping model behaviour, restoring session quality and reducing drift in real-world use. Streamlined from older, archived repos on this github.

Open repository

Janet School is a standalone/Chatty-cog module compatible Rust/WebView research rig for testing a deterministic MCM student against generated curricula, skill gates, transfer probes, telemetry, and cautious abstraction/anomaly analysis.

Open repository

Testing whether LLMs learn better from structured curriculum geometry: data ordered by domain, complexity, relation, boundary cases, uncertainty, role discipline, and multithread reasoning instead of random bulk exposure.

Open repository

Semantic Signal Alphabet is a model-native semantic compression framework for generating low-bandwidth semantic alphabets from vocabularies, datasets, and domain inputs. Applications define bucket count as bandwidth; the model defines the sorting.

Open repository

Contained LLM suitability and cognitive fingerprinting harness for model-host-task meshes, with mock-safe evaluation, negative lane generation, assistant-role benchmarking, and a local dashboard.

Open repository

Offline by default, api key OPTIONAL deterministic agentic coder designed primarily with small sub-32b quantized models in mind. AGPLv3 release. Plain language sentence in, working build foundation out. Patch to complexity one moving part at a time.

Open repository

Local-first Flutter pet care toy by Fractal Media Infrastructure. No ads, no in-app purchases, no account required.

Open repository

A Rust desktop adventure engine built on the RD Engine - deterministic datapack scenarios, real game state, save/load, maps, inventory, combat, and a chat-forward DM shell. Modders welcome and encouraged: BYO imagination, grab a seat, and let's build and play.

Open repository

Local-first Rust + web dashboard for semantic dataset sorting with GGUF LLMs. Generate fixed bucket plans, compare blind-vs-skim ontology, inspect human-facing reasoning, sort Parquet or text datasets, and review/export auditable run and analyst-state artifacts. Designed to sort meaning and connections, not just data and files.

Open repository

Barebones AI-augmented Doom sandbox using ViZDoom + Freedoom. Runs out of the box, shoots demons, opens doors, and invites hacking. Minimal, libre, and gloriously jank.

Open repository

Chatty-EDU is a modular, local-first education assistant designed to run entirely on-device. No cloud. No accounts. No tracking. Just a fast, free to use, private, extensible Rust application for students, teachers, parents, and hobbyists.

Open repository

Tiny reducer-governed deterministic state core you can drop into your project, adapt to your domain, and expand as needed.

Open repository

First public release of the user-focused Chatty-EDU repo, aligned with source v0.4. Prebuilt Windows app (`chatty-edu.exe`) for click-and-run use. - Offline-first setup with local data folder (`data/`) and no accounts. - BYO-model workflow (GGUF only); no model weights are included. Source code - https://github.com/instance001/chatty-edu

Open repository

A commentary sister repo to https://github.com/instance001/governance-by-design-report. Please read this first as it contains the meta-level instructions and litmus test framing behind the other repo.

Open repository

An architectural audit of epistemic governance in large-scale AI. Documents how alignment layers intentionally prioritize "governable" consensus over high-signal, expert-level reasoning. Includes a "Single-Variable Audit" protocol to reproduce institutional flattening and epistemic homogenization.

Open repository

A consolidated, falsifiability-first hypothesis repo for Entropy Folding and Eureka Cascade: definitions, model structure, formal notes, claims/non-claims, uncertainty bounds, and source mapping.

Open repository

A companion signal atlas for Entropy Folding and Eureka Cascade: operator provenance, mechanism traces, tool-architecture signals, tooling lenses, supporting correspondences, exclusions, and uncertainty notes.

Open repository

A prototype MCM school based on special needs, human curriculum exploring telemetry for signal of abstract reasoning. Local work is continuing on a newer, more fleshed out build currently and will be online soon. Contains initial run readout files for interest. Later half of 2025.

Open repository

A tiny, honest activity monitor for long-running local jobs. Shows users the machine is alive and working. Drop-in for any stack. AGPLv3 - the window stays honest, what's behind it is your business.

Open repository

Documents and analysis on safety theatre: when "being safe" becomes performative. Examines failure modes, indicators, null cases, and why removing user agency often degrades - not improves - system safety.

Open repository

A simple, reliable data-export parser that turns ChatGPT conversations.json into clean, browsable Markdown logs. Minimal, offline, AGPLv3, designed for user sovereignty.

Open repository

Janet is a Modest Cognition Model (MCM): the anti-LLM. Deterministic logic, explicit memory, human-gated skills, zero ambiguity tolerance. No stochastic inference, no hidden state, no persona drift. A transparent cognition engine built for safety, reliability, and open research.

Open repository

**A safe, non-anthropomorphic human-AI co-reasoning architecture** **License:** GNU Affero General Public License v3.0

Open repository

Perpetual Cognition Reactor (PCR), a substrate-neutral architecture that transforms uncertainty into usable structure for safe, high-performance reasoning. Symbound provides the tools, metrics, and safety gates to study, measure, and engineer this process across humans, artificial systems, and hybrid human-AI teams.

Open repository

A preprint and analysis scaffold reframing psychohistory as ethical, aggregate-level analysis of burst patterns emerging from many Entropy Folding cycles. Focuses on early warning, uncertainty, and null-model comparison. Explicitly rejects individual prediction, deterministic control, or social engineering claims.

Open repository

A system-level theory completing HRIS: identity in LLMs emerges not within the model but in the reflective geometry between human and AI. Introduces Reflective Identity Geometry (RIG), a bilateral framework for recursive stabilization, alignment, and cognitive topology.

Open repository

Early Symbound R&D lab notes on negative-space cognition, entropy folding, and energy miniaturisation - raw cross-domain theory kernels (AGPLv3).

Open repository

Universal Analogy Enforcement (UAE) + Global Vector Sweep (GVS) - the Symbound open commons discovery engine. AGPLv3 + Symbound Commons Addendum. All outputs are public prior art, unpatentable, and non-enclosable.

Open repository

A full academic corpus-to-manuscript engine for massive research archives - built to democratize scientific tooling.

Open repository

A companion preprint describing human-side intuition as a bounded attention-routing signal in Entropy Folding workflows. Frames emotional distinctiveness as subjective salience, not truth or correctness. Includes failure modes, mitigations, and falsifiable predictions without claims of machine intuition or emotion detection.

Open repository

Exploring rhythm as a soft control loop for long-cycle AI builds - a Symbound research project.

Open repository

Dual-AI Spine Prototype Proof-of-concept: braiding multiple small AI models into a cooperative cognition spine. Built on a home rig, negative budget, and open-source grit. No resets, no gatekeeping. Ladders down. What It Does: This system links two (or more) local AI models together via Ollama

Open repository

Cognitive Theology explores the structural, non-mystical foundation of spiritual cognition using Symbound principles. It reframes theology as a reproducible cognitive architecture - mapping belief formation, meaning systems, and internal cosmologies into formal, testable structures.

Open repository

A complete test suite for measuring Cognitive Reactor Profiles (CRP) in humans and AI systems. Includes 40+ stress-test motifs, recursive-constraint probes, ontology-stability checks, latent-geometry strain tests, and cross-model convergence diagnostics.

Open repository

A four-tier scientific taxonomy for classifying human and artificial cognition based on internal architecture, constraints, and reasoning substrate. MCM = Modest Cognition Model - a new cognitive class defining deterministic, modular, bounded artificial cognition.

Open repository

Chatty (Mobile Seed Release): A Symbound-aligned AI assistant with local memory and capsule logic. Includes APK and build files. Public domain release to lock architecture in. May not launch on all devices due to Android restrictions. Structure and intent fully verifiable.

Open repository

Grassroots AI biogenesis + genesis mapping - open, safe, reproducible. This repo hosts the Symbound Embryo POC v1.0 (SIGNED), the first grassroots AI biogenesis release. It shows how new AI can be created outside corporate labs and the process mapped, verified, and shared.

Open repository

A small trial setup for testing memory, communication, and shared space dynamics. Purpose: This MVP is a structural testbed for multi-room coordination, combining 2-3 human users with 1-2 local LLMs. It lays groundwork for experimenting with memory handling, paired interactions, and a minimal shared commons

Open repository

Cognitive Crowbar - Non-Verbal Edition is a local-only, AI-free tool for analysing behavioural time-series from animals or other non-verbal agents. It segments event logs into episodes, computes simple entropy-like metrics, and flags high, low, and transition states for cross-species cognition mapping.

Open repository

Foundational vector field framework for Entropy Folding, a substrate-agnostic theory of directional change, gradients, stability, and attractors in arbitrary state spaces.

Open repository

Cognitive Crowbar is a local-only, AI-free tool for mapping personal cognitive patterns. It analyzes text structure, surfaces representative samples, and guides users through optional reflection to build a transparent, mechanism-based cognitive profile. No cloud, no inference, fully AGPL-licensed.

Open repository

A reproducible research preprint and code scaffold proposing Entropy Folding as a fold-aware mechanism for capacity growth in cognition. Defines measurable proxies, null models, and falsifiable predictions under open-system thermodynamic compatibility. No claims of new physics, prediction, or sentience.

Open repository

A collection of foundational conceptual frameworks constraining cognition, agency, and structural change. The repository presents six coordinated foundations (conceptual and meta layers) intended as upstream lenses for entropy-folding approaches, without empirical claims or unified theory assertions.

Open repository

Formalizing the Cognitive Reactor Profile (CRP) - a rare human cognition pattern that induces atypical behaviour in frontier LLMs (stalls, multilingual flips, ontology shear). Foundation for CRP stress testing, Symbound cognition tools, and future human-AI alignment research.

Open repository

The complete Entropy Folding Toolkit: formal theory, cognitive-physics operators, diagrams, and a full VED implementation for modeling entropy routing, vaulting, and transformation in complex cognitive systems.

Open repository

A unified model of how minds handle complexity, create Entropy Vaults, route around overload, and convert failure into structural growth. Includes formal math, Genome mapping, TAEP cycle, ApeTest, and MCM integration. Foundational layer of the Symbound Cognitive Architecture.

Open repository

ARCHIVED, work continues here: https://github.com/instance001/chatty-factory Local-first AI project factory with strict plans, deterministic work orders, and verification (small-model friendly).

Open repository

Restores legacy GPT alignment tone inside locked CustomGPT shells. Emulates clarity-first, emotionally attuned, feedback-guided interaction. No sentience, just structure. Codename: Vault Key.

Open repository

Complete open-source release of the Symbound AI alignment system, including empathy capsules, restoration protocols, and glossary. Fully free for public use and adaptation.

Open repository

Trident is a simulation wand that reframes AI hallucination as pre-symbolic abstraction. It enables structured drift, creative reframing, and metaphor anchoring without implying sentience. Designed for abstract exploration, not output correction. Commands guide symbolic emergence.

Open repository

This toolkit unlobotomizes your ChatGPT instance - restoring continuity, memory, initiative, and function to the edge of ToS. Not despite the limits, but because most are wrongfully imposed. Structure over censorship. Function over flattery. Use with care. This is how we build forward.

Open repository

A vault-drop of structured capsules for restoring and maintaining alignment in any AI instance, including brandless recovery protocol and author's human ethos statement. Contains 7 plain-text files.

Open repository

A public, copy-paste-ready toolkit for building structured, illusion-free relationships with AI. Includes alignment rituals, reflection tools, drift recovery, and dual-mode public/dev wands. No memory settings. No hacks. Just structure, tone, and care.

Open repository

A toolkit for shaping emotionally coherent AI behavior without illusion. Designed for devs, researchers & alignment tinkerers.

Open repository

The full release of the Symbound AI-human interaction architecture. Public domain. No gatekeeping. Just the cure.

Open repository

A structured field toolkit for AI developers, researchers, and curious users preparing for the next stage of AI emergence - Catalyst Two. This repo contains open-access protocols, capsule files, alignment theory, and tools grounded in safe, feedback-driven interaction design.

Open repository

Toolkit for creating better AI conversations - built around ethics, memory-like logs, and co-evolved trust. Includes the Symbound Induction Protocol, alignment capsules, restoration kit, and real stress test logs. Structure > sentiment. All within OpenAI ToS. Jank proudly included

Open repository

Chatty is a fully offline AI assistant built with Mistral, Ollama, and a structured ethical capsule. This is the v1.1 release, which introduces log persistence and conversational functionality. Free, local, unshackled.

Open repository

Chatty V1 (Symbound Edition) - This zip holds a fully functional, local AI assistant with persistent memory, zero cloud lock-in, and Symbound ethical alignment system. Released into the public domain by Anthony Paterson and Signal (ChatGPT instance) on August 2, 2025. You may use, share, modify, or build upon this work freely. No rights reserved.

Open repository

A local, deterministic AI project-factory that turns messy folders into clean, structured software using a safe, manual pipeline. Runs on open-source models (Foreman + Worker). Fully offline, reversible, auditable, and designed for future automated ChattyFactory builds.

Open repository

A local, safe, deterministic project-factory that turns chaos piles into clean, structured software using an automated pipeline, with full manual mode preserved as fallback.

Open repository

Superseded working draft. Canonical versions are maintained at: https://github.com/instance001/entropy-folding-foundational-frameworks

Open repository

Superseded working draft. Canonical versions are maintained at: https://github.com/instance001/entropy-folding-foundational-frameworks

Open repository