Memory & Knowledge MCP Servers

Memory & Knowledge servers solve the forgetting problem. They give an agent somewhere to put what it learns and a way to get it back: vector stores for semantic search, knowledge graphs for relationships, document indexes for your own material, and sync layers for note apps and wikis.

This category pairs well with everything else. An agent that remembers decisions, project context, and past research across sessions is noticeably better at long running work than one that starts cold every time.

The live list did not load just now. The full category is still available in search.

Browse Memory & Knowledge servers in search

Questions about Memory & Knowledge servers

What is the difference between memory and RAG servers?

Memory servers store what an agent learns at runtime. RAG style servers index documents you provide. Many packages do both, and the package page spells out which.

Where does my data live?

Wherever the server stores it: locally for stdio servers with local indexes, or in whatever backend you configure. Nothing is uploaded to Pharos.

Which ones are most used?

The download sorted top 24 is above, and the full category is searchable with capability filters.