brainary 0.0.1

A cognitive architecture framework integrating LLMs, capabilities, and self-regulation.

Brainary + PoK

A brain-inspired computing architecture (Brainary) and a language for knowledge programming (PoK).

Workflow

┌───────────────────────────┐ │ Python Script │ │ (calls Brainary APIs) │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Brainary VM │ │ accept_op(ActionOp) │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Scheduler │ │ _estimate(op) │ │ ─ Determine relevant │ │ capabilities (CT, │ │ Planning, Reasoning, │ │ Evaluation, etc.) │ │ ─ Select strategies via │ │ Knowledge + Experience │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Apply Critical Thinking│ │ (pre-analysis / BVCA) │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ ActionOp.render │ │ - instruction │ │ - contexts │ │ - pre-analysis │ │ - applied strategies │ │ - arguments │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Problem Solving Module │ │ (LLM execution) │ │ → produces result │ └─────────────┬─────────────┘ │ ▼ ┌───────────────────────────┐ │ Monitor │ │ estimate_reward(op) │ │ capability-aware scoring │ └─────────────┬─────────────┘ │ Reward ≥ Expected? ┌─────────────┐ │ Yes │ ▼ │ ┌─────────────┐ │ │ Record │ │ │ execution │ │ │ & strategies│ │ └─────────────┘ │ │ ┌─────────────┴─────────────┐ │ No (reward < expected) │ ▼ │ ┌───────────────────────────┐ │ │ Scheduler Replanning │◄─────┘ │ - Feedback incorporated │ │ - Re-select strategies │ │ using Knowledge + LLM │ └─────────────┬─────────────┘ │ ▼ ┌─────────────────────┐ │ Loop to Execution │ └─────────────────────┘

Planning

brainary/capabilities/planning/ ├── planning_base.py # Base Planning class │ ├── # --- Cognitive / Human-like --- ├── hierarchical_planning.py # Break down goals into subgoals (generalized HTN-style) ├── forward_planning.py # Start from initial state, simulate toward goal ├── backward_planning.py # Start from goal, work backwards to requirements ├── contingency_planning.py # Plan for “what if” scenarios under uncertainty ├── opportunistic_planning.py # Adjust plan when new opportunities arise ├── adaptive_planning.py # Revise plan dynamically from feedback │ ├── # --- Algorithmic / Formal AI --- ├── htn_planning.py # Hierarchical Task Networks (formalized decomposition) ├── means_end_planning.py # Resolve gaps between current and goal states ├── critical_path_planning.py # Optimize for dependencies & time constraints ├── mcts_planning.py # Monte Carlo Tree Search, explore via simulation ├── default_planning.py # Fallback strategy