NeuralMemory stores experiences as interconnected neurons and recalls them through spreading activation, mimicking how the human brain works. Instead of searching a database, memories are retrieved...
Does this issue happen to you?
I built NeuralMemory to fix this problem.
Install it → The AI will remember decisions, patterns, and bugs that have been solved.
Open source, free. Please help me test it! 🙏
Neural Memory: Teaching AI to remember like we think. 🧠 A digital brain system with a spreading activation mechanism and intelligent memory management. It has been highly appreciated by the Unikorn community and continuously updated to lead the trend of Agentic Workflow.
"Neural Memory is not just an ordinary database; it is a memory system inspired by the human brain itself. Instead of merely storing static data, Neural Memory transforms all information into tightly connected 'neurons' through the Spreading Activation mechanism. Notably, the Gentle Forgetting feature helps your AI maintain sharpness by removing junk data, retaining only the most valuable 'memories'."
👉 pip install neural-memory nmem init --full
Domain boundaries (B5 Phase 3): Scope boundary memories to specific domains (e.g. , , ). Boundaries without a domain remain global. Uses existing tag system with prefix convention — zero schema migration. financialsecuritycode-reviewdomain:
nmem_remember(content="...", type="boundary", domain="financial") → auto-adds tagdomain:financial
nmem_recall(domain="financial") → HOT context injection filters boundaries by domain, keeps global (unscoped) boundaries
nmem_boundaries tool — list boundaries grouped by domain, list unique domains with counts
Brain milestone tool (): Track neuron-count achievements and generate growth reportsnmem_milestone
Improved
MCP tool count: 53 → 55 (added , nmem_boundariesnmem_milestone)
Auto-tier engine (B5 Phase 1, Pro): Automatic WARM→HOT promotion, HOT→WARM demotion, WARM→COLD archival based on access patterns. Protection for BOUNDARY types and pinned fibers. Oscillation prevention. MCP tool with status/evaluate/apply/history/config actions.nmem_tier
Decision intelligence (B5 Phase 2): Extract structured decision components (chosen, alternatives, reasoning, confidence) from DECISION-type memories. Detect overlapping prior decisions, classify relationships (confirms/contradicts/evolves), create EVOLVES_FROM synapses, boost recall scores for domain-relevant decisions.
Dashboard Phosphor icons: Migrated all 19 component files from Lucide to Phosphor Icons (). Added Playwright E2E smoke tests (8 tests).@phosphor-icons/react
Have you used this product? Leave a review!
Chat AI Nhập vai dành cho người Việt
AI Agent Capability Expansion Toolkit
Your AI Development & Marketing Teams
Chia sẻ và khám phá sản phẩm do người Việt xây dựng
Bộ công cụ tạo AI Skill từ ý tưởng — Dành cho Antigravity, Claude Code
Trợ lý dinh dưỡng thông minh
No one has supported Neural Memory yet.
Reviews & comments