mcp-server-memory
Mirror of
MCP-Mirror
README
mcp-server-memory
This is an MCP server to interact with a memory text file to help Claude with inter-chat context.
Each line is a memory.
These tools allow Claude (and other MCP clients) to manage memories mid-chat:
memory_add(memory: string)
- append the memorymemory_search(query: string)
- return matching memories (substring exact match) - later, might allow globs/regexmemory_delete(query: string)
- delete matching memories (substring exact match)memory_list()
- return all memories- FYI
memory_update
==memory_delete
+memory_add
For example,
- I mention my name => "talking to Wes"
- metion daughter's age => "Wes's daughter is 8"
- say working on a typescript project => "working on typescript project"
- AND, this is critical, can be based on things Claude (assistant/LLM) says or does...
- Notably, tool use (i.e.
run_command
)... say there is a failure on a first attempt to use the tool (i.e. thepython
command isn't present) and then a subsequent tool use succeeds (i.e. usingpython3
instead ofpython
) => Claude can record "use python3, python is not present"...
- Notably, tool use (i.e.
- I ask Claude to get rid of memories about X => memory_delete(query: X)
- I correct my name => memory_search("oldname") + memory_delete(each matching record, or a common subset query) + memory_add("newname")
Then, when a new chat begins, Claude will automatically get recent memories (a subset or all) OR can ask for memories (some/more/all). And then can use those to influence responses/tools/etc.
Design
A simple memory text file, why:
- ChatGPT's memory works well and is essentially a text file
- Maybe it's structured behind the scenes, however if you review your memory its presented as a text file.
- My testing of a similar reminders feature for
mcp-server-commands
worked great (when Claude had them). - Unstructured text simplifies the tooling and parameters to basically managing a list of strings.
Cueing mechanism:
- It's also important to have a cue for the model to know when to store memories. This is a bit more unclear how best to do this but..
- Training: OpenAI acknowledges some training of models to know when to store memories. Just like models are trained for tool use.
- Prompt: A system prompt component likely contains a reminder to trigger storing memories.
- Tool alone: In my testing of Claude, with a tool spec alone, and even with hints/suggestions in tool responses, I couldn't get Claude to store memories. So this alone is not sufficient. Seems like Claude's training with tools is to only use them in pursuit of the prompt/request and thus why I believe adding a reminder/cue in a prompt component will work well.
TODOs/Ideas
I have no idea if these are worth the time, just listing ideas here for the future. Perhaps in part to stop myself from working on them :)
- Recency factor: a way to rearrange memories based on recency?
- Order then becomes relevant for ambiguous memory queries (i.e. work on typescript project and python project then I ask to start a new project, could suggest the most recently used one?)
- Fade out old memories?
Recommended Servers
Crypto Price & Market Analysis MCP Server
A Model Context Protocol (MCP) server that provides comprehensive cryptocurrency analysis using the CoinCap API. This server offers real-time price data, market analysis, and historical trends through an easy-to-use interface.
MCP PubMed Search
Server to search PubMed (PubMed is a free, online database that allows users to search for biomedical and life sciences literature). I have created on a day MCP came out but was on vacation, I saw someone post similar server in your DB, but figured to post mine.
dbt Semantic Layer MCP Server
A server that enables querying the dbt Semantic Layer through natural language conversations with Claude Desktop and other AI assistants, allowing users to discover metrics, create queries, analyze data, and visualize results.
mixpanel
Connect to your Mixpanel data. Query events, retention, and funnel data from Mixpanel analytics.

Sequential Thinking MCP Server
This server facilitates structured problem-solving by breaking down complex issues into sequential steps, supporting revisions, and enabling multiple solution paths through full MCP integration.

Nefino MCP Server
Provides large language models with access to news and information about renewable energy projects in Germany, allowing filtering by location, topic (solar, wind, hydrogen), and date range.
Vectorize
Vectorize MCP server for advanced retrieval, Private Deep Research, Anything-to-Markdown file extraction and text chunking.
Mathematica Documentation MCP server
A server that provides access to Mathematica documentation through FastMCP, enabling users to retrieve function documentation and list package symbols from Wolfram Mathematica.
kb-mcp-server
An MCP server aimed to be portable, local, easy and convenient to support semantic/graph based retrieval of txtai "all in one" embeddings database. Any txtai embeddings db in tar.gz form can be loaded
Research MCP Server
The server functions as an MCP server to interact with Notion for retrieving and creating survey data, integrating with the Claude Desktop Client for conducting and reviewing surveys.