Projects

Theory made executable.

Each project is a test of the Lucid framework against real conditions. Not demos — working tools producing real signal.

Active

Working prototype

Lucid Decision System

A personal decision-reasoning system that tracks decisions as epistemic objects, measures prediction accuracy over time, and accumulates structured learning from every outcome. Built on PostgreSQL + Qdrant, with a local embedding pipeline and a Retrieval Broker that surfaces relevant past decisions when new ones are started.

Epistemic Store

PostgreSQL — decisions, outcomes, assumptions, relationships

Semantic Index

Qdrant + nomic-embed-text — 768-dim local embeddings

Learning Engine

Assumption history, DQS scoring, capability pressure thresholds

Retrieval Broker

FastAPI — similarity × 0.30 + ECS × 0.40 + recency × 0.20

The decision loop is a working prototype; its evidence base is published in the repository.

Build Roadmap

Four phases. Build-first. Theory is event-driven.

✓ 0

Schema Foundation

Postgres running locally. Core tables migrated. Seeded with one real past decision.

✓ 1

Decision Loop

FastAPI + CLI. Record a decision, record an outcome, get DQS back.

✓ 2

Learning Engine

Assumption history, capability pressure thresholds, perspective score updates.

✓ 3

Memory Layer

Qdrant + local embeddings + Retrieval Broker. New decisions surface relevant past ones.

✓ 4

Web Presence

thelucidmind.ai and mykungfu.ai as real, navigable experiences.

View system architecture →|Systems overview →