A Pre-Thinker For Neuro-Symbolic Knowledge Capture.

This hub is the fast path to the project spine: Pre-Thinker control plane, write-path safeguards, live transcripts, and benchmark artifacts. Use the demo and archive sections below as the running trail of where the project has been and where it is heading next.

Read Pre-Thinker Control Plane Open Flagship Demo Agent Fast Start View Repository Research Archive

demos quickstart core docs research html index

Live Transcripts

public rendered sessions

MCP Tool Surface

tested in surface playbook

Tests Passing

115

current documented baseline

Mode

local-first

write-path reliability lane

Live Demos

Rendered sessions are where the system is easiest to judge: tool calls, responses, and behavior in context.

flagship

Hospital Critical Path Method (CPM) Control Room

Dependency propagation, supplier incident handling, and recovery updates in a deterministic operations scenario.

Open Interactive Transcript Read Markdown View JSON Capture

simulation

Fantasy Overlord Session

Pause/resume scenario edits, state updates, and multi-hop effects shown in an operational simulation flow.

Open Transcript Walkthrough

launch ops

Indie Launch War Room

Conversational launch-ops flow with deterministic risk propagation and uncertain-note intake handling.

Open Transcript Walkthrough

tooling stress

Natural Language Robustness Session

Honest pass/miss capture for tool selection under more natural phrasing.

Open Transcript Read Notes

surface check

MCP Surface Playbook Session

Compact run that exercises the full MCP tool surface and captures expected behavior markers.

Open Transcript Playbook Artifact

memory ladder

Family Memory Ladder (Hybrid Tracks)

Harder ad-hoc ladder with corrections, tentative notes, branch rewiring, and context pressure in publishable conversational captures.

Open Hybrid Transcript Open Hybrid Pressure Transcript Read Report

Quickstart

Use this if you want to go from clone to a local MCP-backed demo quickly.

install + verify

Run The Baseline

Install deps, run tests, and sanity-check docs consistency before demo capture work.

git clone <repo>
cd prolog-reasoning
pip install -r requirements.txt
python -m pytest tests -q
python scripts/check_docs_consistency.py

mcp setup

Connect Local LLM Tools

Follow LM Studio MCP setup, then run chat playbooks or scripted captures.

LM Studio MCP Guide MCP Chat Playbooks Walkthrough Ladder

Read First

Load-bearing docs for the current trajectory: pre-thinker scope, write-path policy, and runtime reality.

Project Spine

Pre-Thinker + Write Path

Unified Intake / Memory / Write Spec
Agent Ingestion Tests
Pre-Thinker Control Plane
Local-first pre-think can handle some low-complexity control turns before cloud escalation, which may reduce token-credit spend.
Latest Pre-Thinker Infographic

Walkthroughs

Research And Benchmarks

Model comparison and ingestion-quality work focused on failure legibility, safe abstention, and write-discipline reliability.

Model Matrices

Collaboration + Direction

Complete HTML Index

All currently published HTML pages under docs/, including research renders and run artifacts.

Hub And Root Pages

Examples Pages

Research Ledgers

Research Run Archives

Public pages root: https://dr3d.github.io/prolog-reasoning/
If GitHub Pages source is main / docs, this hub lives at /docs-hub.html.