TS-Core

Status: flagship.

Canonical route: TS-Start-Here -> TS-Reasoner-v0 -> TensionLM -> TS-Codex-OS / TS-Core / CIG.

Role in the TS stack: Lightweight graph/tension runtime kernel for TS-style propagation, relaxation, stability, and inspectable constraint experiments.

What this repo is: A local-first graph engine and CLI for wave propagation, activation/stability dynamics, weighted edges, persistence, and tension telemetry.

What this repo is not: Not a full reasoning model, not an LLM, and not evidence of general intelligence by itself.

Start here: install the package, run the local runtime, then inspect the propagation examples and tests.

Thinking System Graph Dynamics Kernel

Lightweight, local-first graph engine with wave propagation for the TS (Thinking System) architecture.

Python-first with optional Rust acceleration (PyO3). Designed as the reusable core for wave-based cognitive systems.

---

Quick Start

git clone https://github.com/BoggersTheFish/TS-Core.git
cd TS-Core

python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"

# Run the TUI / mind runtime
python -m src.python.mind_runtime

Or use the daily grounding spin:

python -m src.python.daily_spin

---

What It Does

• Maintains a graph of nodes (with activation + stability) and weighted edges
• Provides a typed tension kernel surface: GraphState, TensionChannel, ChannelResult, ResolverEvent, RuntimeTrace, and Receipt
• Runs wave propagation cycles (standard + Kernel Wave 12 mode)
• Applies filter layers and narrative/Icarus voice overlays
• Persistent JSON + JSONL history
• Optional Z3 constraint solver for toy alignment checks
• Optional Grok/xAI plugin

---

Typed Tension Kernel

The typed kernel lives in the importable ts_core package. It is intentionally domain-neutral:

GraphState -> TensionChannel[] -> TypedTensionRuntime -> RuntimeTrace

Run the generic demo:

python examples/typed_tension_kernel_demo.py

This demo does not use reasoning-specific English or syllogism logic. App layers such as TS-Reasoner define their own channels on top of the kernel.

---

Key Modes

• Standard wave propagation
• Kernel Wave 12 — advanced 9-phase strongest-node biased cycle
• Daily Spin — quiet grounding + lowest-stability push

---

Part of the TS Ecosystem

• Powers graph dynamics for BoggersTheAI and BoggersTheCIG
• Foundation for GOAT-TS experiments
• Used in TensionLM training loops

---

Philosophy

This repo implements the mechanical substrate of TS: constraint graphs + continuous wave dynamics that produce emergent stability.

The narrative and philosophical layers sit on top of the actual graph engine.

---

Default branch: master

License: MIT

Status: Stable kernel, actively used across the TS project