Skip to content

Release v0.1.0 - Swarm Visualization

Choose a tag to compare

View all tags
@anonimity69 anonimity69 released this 20 Apr 00:48
· 94 commits to main since this release
v0.1.0
a8ba2d9

Release v0.1.0: Swarm Visualization & Architecture Clarity

Welcome to the v0.1.0 release of GraphCortex! This release focuses on making the complex neuro-symbolic memory architecture transparent and accessible through high-fidelity visual documentation and refined system flows.

🚀 Key Highlights

🧠 Swarm Intelligence Visualization

This release introduces formal Mermaid architecture diagrams natively in our documentation.

  • System Architecture: Visualize the interaction between the Swarm Engine agents (Researcher, Summarizer, Librarian), the Neo4j graph store, and the RL policy backend.
  • Data Flow Lifecycle: A sequence-level view of how a user query triggers Spreading Activation and how the background Librarian agent identifies and "self-heals" redundant knowledge.

🛡️ Self-Healing Memory (RL Curation)

We've refined the Librarian agent's RL policy integration:

  • Autonomous Merge/Prune: The policy now better identifies "Graph Heat" signals to collapse redundant nodes and prune stale context.
  • Error Sanitization: Improved heuristics for cleaning up extraction noise and rate-limit artifacts from existing memories.

🔍 Dual-Trigger Retrieval

The RetrievalEngine now robustly supports Hybrid Anchor Search:

  • Combines BM25 Lexical matching with Dense Vector Semantic search to ensure anchors are found even with imprecise queries.
  • Optimized Lateral Inhibition logic to prevent "Hub Explosion" during graph traversal.

🛠️ What's Changed

  • Documentation: Overhaul of README.md with structured Architecture and Data Flow sections.
  • Git Workflow: Resolved divergent branch issues and synchronized documentation across the swarm.
  • Core Engine: Finalized interface bindings for the Librarian's curation loop.

📥 Quick Install

git clone https://github.com/anonimity69/GraphCortex.git
cd GraphCortex
./setup.sh

🌌 The Vision

GraphCortex is built on the premise that agent memory should be a first-class, self-improving system. This release is a major step toward making that "intelligence layer" readable and reliable for long-running autonomous deployments.

Built for agents that need to think longer than one conversation.

Assets 2
Loading