cognitive-systems-engineering

What problem does it solve?

This Skill helps architects and engineers move beyond stateless prompt-response agents by providing blueprints for persistent, goal-directed AI minds that manage memory, make repeatable decisions, and coordinate multiple specialist agents.

Core Features & Use Cases

• Tiered Memory Design: Patterns for Working, Episodic, Semantic, and Procedural memory using sliding windows, vector stores (RAG), and context files.
• Decision Cycle Implementations: Guidance for choosing and implementing OODA, ReAct, or production decision loops with retry counters and escalation.
• Multi-Agent Patterns: Orchestrator-workers, blackboard, pipeline, and debate topologies for decomposing complex tasks across specialists.
• Use Case: Build a customer-support agent that retains session context, retrieves past interactions via embeddings, chooses tools to act, and delegates billing tasks to a payment worker.

Quick Start

Design a cognitive agent architecture with working, episodic, semantic, and procedural memory, select an OODA or ReAct decision cycle, and propose a multi-agent pattern for coordinating tool use and retries.