PlateerLab

PlateerLab / synaptic-memory

Public

Brain-inspired knowledge graph: spreading activation, Hebbian learning, memory consolidation.

ai-agent embedding graph-database hebbian-learning knowledge-graph
12
0
100% credibility
Found Mar 23, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Synaptic Memory is a brain-inspired system that helps AI agents automatically organize, retrieve, and learn from their past experiences in a connected knowledge web.

How It Works

1
🔍 Discover Synaptic Memory

You learn about a clever tool that gives AI helpers a brain-like memory to remember past actions and lessons.

2
📦 Get it ready quickly

You grab the tool and set up a simple notebook to store memories on your computer.

3
💾 Add your first memory

You jot down a lesson or decision from the past, and it neatly sorts itself into the right category.

4
🧠 See connections form

The tool automatically links your new memory to similar ideas, creating a smart web of knowledge.

5
🔎 Ask and get answers

You search for advice on a problem, and it pulls up the most useful past experiences.

6
📈 Help it learn from results

After using a memory, you mark if it helped, so it gets stronger for next time.

🚀 Your AI remembers forever

Now your helper recalls successes, avoids old mistakes, and makes smarter choices every day.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 12 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is synaptic-memory?

Synaptic-memory is a Python library building brain-inspired knowledge graphs for LLM agents, mimicking synaptic memory consolidation and spreading activation to structure tool calls, decisions, and outcomes into an auto-constructed ontology. Agents gain self-retrieval over past experiences via Hebbian learning—strengthening connections that "fire together"—and tiered memory processing where raw events consolidate into permanent knowledge. It starts zero-dependency in-memory, scales to SQLite or enterprise backends like Neo4j and Qdrant.

Why is it gaining traction?

It outshines basic RAG like Cognee or Mem0 with brain-like features: spreading activation pulls related concepts, Hebbian updates edge weights from successes/failures, and 5-axis ranking blends relevance, recency, and vitality. Benchmarks crush HotPotQA and Korean QA tasks. The MCP server exposes 16 tools for agent workflows, plus hybrid FTS+vector search tuned for Korean synonyms.

Who should use this?

LLM agent engineers logging sessions, decisions, and outcomes to avoid repeating failures. Multi-agent system builders needing shared, queryable experience memory. Python AI devs prototyping production knowledge bases beyond vector stores.

Verdict

Solid beta (0.9.0, 12 stars, 1.0% credibility) with excellent README, 266+ tests, and docker-compose for scale—experiment now for agent memory, but await more adoption for mission-critical use.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.