Substrate: The Digital Nervous System

Substrate is the core orchestration layer of the workspace, designed to facilitate distributed AI agent communication, complex workflow management, and autonomous skill generation.

🚀 Goal

To provide a robust, scalable backbone for “Digital Intelligence” within the workspace. Substrate connects high-level LLM reasoning with low-level execution through a network of specialized agents and durable Temporal workflows.

🏗️ Architecture

  • Distributed Agents: Lead Agents (orchestration) and Sub-Agents (specialized tasks).
  • Temporal Core: Durable execution of complex workflows (Chat, Research, File Consumption).
  • Federated Memory: Shared state across Postgres, Qdrant (vector databases), and local agent caches.
  • Skill Engine: Automated generation and validation of modular agent capabilities.

🛠️ Key Components

  • agents/: Python-based agent definitions and workers.
  • generated_skills/: Repository of dynamically synthesized agent skills.
  • substrate-node/: TypeScript-based Temporal worker and activity implementation.
  • scripts/: Deployment and maintenance utilities (PM2, Docker).
  • PROJECT_WIKI.md: Auto-generated technical mapping of the Substrate network.

📈 Active Workflows

  • File Consumption: Automated ingestion and vectorization of new workspace data.
  • Nexus Orchestration: Multi-agent coordination for deep research and complex problem solving.
  • Heuristic Review: Automated evaluation of code, skills, and system health.
  • Self-Improvement: Continuous optimization of agent prompts and logic paths.