D3-TUI LLM Wiki Index

Start Here

• Schema
• Agent instructions
• Architecture research queue
• Research answer template
• Task lifecycle
• Claim board
• Agent roles
• KOS contract placeholder
• Flycast smoke placeholder

Runs

New runs should create runs/<run-id>/task-clock.md and append to log.

Architecture

• [[langgraph-engine-evaluation]] — Pygame engine survey: Wargame Engine recommended, Sequtus as reference, 7 engines evaluated

• [[langgraph-layered-questions]] — 27 open questions across Pi (7), Game (10), and Membrane (10) layers

• [[langgraph-unit-node-interaction-model]] — Units as state cases, nodes as task spaces, access gating, config overrides, passive intel harvesting

• [[langgraph-tool-execution-tradeoffs]] — LangChain tools vs external agent harnesses inside LangGraph nodes

• [[langgraph-node-anatomy]] — Full catalog of what a LangGraph node can contain (input, process, output, references, constraints)

• [[langgraph-hex-node-mapping]] — Hex/tile to LangGraph node mapping concept exploration

• [[langgraph-gameplay-modes]] — Auto-run vs control mode + per-unit/lane AI toggles

• [[langgraph-game-surface]] — LangGraph as turn-based strategy game backend + console client architecture (RG40XXV/Wii/PS2 → Pi)