memcord_nameLABSCO SUMMARY
Memory you can point at a folder.
The part that makes this usable day to day is the binding file: drop a .memcord in a project and every save and read routes to that project's slot, so you stop naming things and start just talking.
What it is
A self-hosted MCP server that organises chat history into named memory slots, summarises it, and searches across everything you have saved. Storage is local, and slots can be bound to a project directory so the assistant finds the right one without being told.
What you get
- 23 tools over memory slots:
memcord_name,memcord_use,memcord_save,memcord_read,memcord_list,memcord_ping memcord_save_progresssummarises before saving, with a compression ratio you choose- Search and questions across saved chats:
memcord_searchfor full text,memcord_queryfor a natural-language question,memcord_select_entryto jump to a point in the timeline - Organisation that survives growth:
memcord_tag,memcord_list_tags,memcord_group,memcord_mergefor combining related slots,memcord_compressandmemcord_archive memcord_importto bring outside material in,memcord_exportandmemcord_shareto get it back outmemcord_zeroandmemcord_unbindwhen you want a slot or a binding gone- Slot resolution that follows a fixed order — an explicit
slot_name, then the active slot, then the.memcordfile in the working directory - Four summarizer backends switchable per slot without a restart:
nltk,sumy,semantic,transformers - A slot's data file pointed at any directory through
memcord_configure, so a shared folder carries both the data and the slot's settings between machines
Requirements
Python 3.10+ and uv; the installer handles both and writes the client configuration for Claude Code, Claude Desktop, VSCode and Antigravity. Fresh installs register globally in ~/.claude.json; an older project-scoped install stays that way until you rerun the config script with --scope user. Nothing leaves the machine — there is no cloud side. Version 4.3.6, MIT.
Setup effort
One command — curl -fsSL https://github.com/ukkit/memcord/raw/main/install.sh | bash
USE CASES
01
Picking a project back up after a week away
The slot bound to the directory is found and activated on its own, and reading it back replays where the work stopped.
02
Keeping a long session without keeping every word
Save with summarisation instead of raw text, and set how hard it compresses.
03
Finding a decision you know you made somewhere
Full-text search across slots, or ask the question directly and let it answer from what was saved.
04
Collapsing three overlapping conversations into one record
Merge the slots and the duplicated ground gets folded together rather than repeated.
05
Sharing one project's memory across two machines
Point the slot's storage at a synced folder; the settings travel with the data, so both machines see the same configuration.
06
Getting a record out to somewhere else
Export a slot in the format you need, or bring outside notes in and index them alongside the chats.
23 tools
Listed in project docs. Your client may show a different list.
memcord_usememcord_savememcord_readmemcord_save_progressmemcord_configurememcord_listmemcord_pingmemcord_searchmemcord_querymemcord_zeromemcord_select_entrymemcord_mergememcord_initmemcord_unbindmemcord_compressmemcord_tagmemcord_list_tagsmemcord_groupmemcord_importmemcord_archivememcord_exportmemcord_shareBEFORE YOU INSTALL
What to know first
Tools you get
23 tools named in the project's documentation.
Requires
Python 3.10+
INSTALLATION
Connect your client
Server setup
Setup
TERMINAL
git clone https://github.com/ukkit/memcord.git
cd memcord
uv venv && uv pip install -e .
uv run python scripts/generate-config.pyOr run the install script
TERMINAL
curl -fsSL https://github.com/ukkit/memcord/raw/main/install.sh | bashMCP Clients
Claude Code
CLAUDE CODE · TERMINAL
claude mcp add memcord uv --directory <YOUR_PATH_TO_MEMCORD> run memcord --scope user -e PYTHONPATH=<YOUR_PATH_TO_MEMCORD>/srcClaude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"memcord": {
"command": "uv",
"args": [
"--directory",
"<YOUR_PATH_TO_MEMCORD>",
"run",
"memcord"
],
"env": {
"PYTHONPATH": "<YOUR_PATH_TO_MEMCORD>/src"
}
}
}
}VS Code
VS CODE · JSON
{
"servers": {
"memcord": {
"command": "uv",
"args": [
"--directory",
"<YOUR_PATH_TO_MEMCORD>",
"run",
"memcord"
],
"env": {
"PYTHONPATH": "<YOUR_PATH_TO_MEMCORD>/src"
}
}
}
}OpenClaw
OPENCLAW · TERMINAL
openclaw skills install memcord --force
