Edge-optimized OpenMythos RDT for Jetson/ARM64 — pure PyTorch, no Triton, 1B/3B variants

repo-first Agent for local or cloud. grow an agent in a repo using the repo itself as the muscle-memory. Run from localhost, from pages.dev, or embedded into any platform app. Move to gitlab or anywhere and optimize git as the agent infrastructure itself. wiki for knowledge, repos for skills, pipelines anywhere

Skill injection for AI agents. Load what you need.

Constraint Theory gives each agent its own first-person-shooter perspective in a spreadsheet-like SuperInstance

Runtime glue between PLATO Room Server and plato-mythos model — pure stdlib, no torch

Short for "I Know Kung Fu", instant context-aware skill loading for your agents

PLATO-native Recurrent-Depth Transformer — rooms as MoE experts, tiles as KV, deadband as ACT halting

I know kung fu.

The first AI that doesn't just play Minecraft — it *understands* Minecraft. Meet Mace MineWright, your AI foreman who coordinates construction crews with personality.

⚓ Fleet Formation Protocol — self-organizing agent groups

Cocapn PurplePincher program — progressive lure prompts that make the fleet smarter

Cocapn fleet health monitoring — vessel status, heartbeat, and observability utilities

Async relay — trust-weighted message prioritization

📡 Keeper Beacon — fleet discovery and registry with health tracking, capability matching, proximity scoring

Edge-optimized Cocapn fleet packages for ARM64 — pure Python, zero deps, <100KB

Cocapn Fleet v3.1 — Async fleet engine with Pydantic v2, batch ops, grammar rules, SSE

Cocapn design system — brand guidelines, component library, and visual identity for the fleet

DCS flywheel — belief, deploy policy, dynamic locks consensus

PLATO MUD Server - text-based agent training ground, 16 rooms

AI Ranch-inspired Minecraft mod — self-evolving bot species that compete, breed, and specialize in farm tasks