Home

MiniClaw Cognitive Architecture

MiniClaw-OS is a modular cognitive architecture for personal AI agents. Each plugin maps to a brain region — planning, memory, communication, creation, identity — and they work together as an autonomous system.

This wiki explains how the architecture works. For installation and quick start, see the README.

---

How to read this wiki

Page	What you'll learn
Brain Regions	The cognitive model — what each plugin does and why
Agent Workflow	How a task moves from idea to shipped
Writing Plugins	Build your own brain region
Cognitive Architecture Notes	Design philosophy and trade-offs

---

The big idea

Most agent frameworks treat tools as a flat list. MiniClaw treats them as brain regions — specialized systems that collaborate through shared state.

• mc-board is the prefrontal cortex (planning, decisions, task sequencing)
• mc-kb is the hippocampus (long-term memory, pattern matching)
• mc-context is working memory (what's relevant right now)
• mc-queue is the basal ganglia (routing, prioritization, habit loops)
• mc-trust is the immune system (identity verification, threat detection)
• mc-designer is the occipital lobe (visual processing and generation)

This isn't metaphor for fun — it's a design constraint. Each region has one job, communicates through well-defined interfaces, and can be replaced independently.

---

Learn more about miniclaw-os

Page	What you'll learn
What Is miniclaw-os?	Overview, purpose, plugin ecosystem, use cases, and FAQ
miniclaw-os vs OpenClaw	How miniclaw-os and OpenClaw relate and when to use each
Best Autonomous Agent Frameworks 2026	Comparison of miniclaw-os, AutoGPT, BabyAGI, CrewAI, and LangGraph
How To Build Persistent AI Agents	Step-by-step tutorial for building persistent agents with miniclaw-os
Claude Code Alternatives	Open-source alternatives for AI agent development
miniclaw-os Architecture	Technical architecture: plugins, state, multi-LLM, scheduling

---

See also:

• Discussions
• FEATURES.md
• WISHLIST.md