ori-memory — Claude Code MCP server: memory and setup
Keeps agent memory available across sessions, clients, and machines, with search over linked knowledge. Useful when project context needs to outlast a single chat.
claude mcp add ori-memory -- npx -y ori-memoryWhat ori-memory does
ori-memory stores information in Markdown files, connects notes with wiki-links, and uses Git for version control. An index is built from those files; the MCP server gives an agent access to it, including through the ori_recall tool. The same index can also be used from the CLI or the library.
Add the server with claude mcp add ori-memory -- npx -y ori-memory. To create an index from the CLI, run npx ori init and npx ori index build. Markdown files hold the underlying information, but learned retrieval state is stored separately, so export it before deleting data.
What it helps with
- keep agent memory across sessions
- find related facts in a knowledge graph
- connect shared memory to an MCP client
- build an index from Markdown files
Example prompts for the agent
- Remember this decision about caching so I can retrieve it in a later session.
- Find what we decided about caching in the project memory.
- Show notes linked to the architecture decision.
How to install ori-memory
Platform CLI
claude mcp add ori-memory -- npx -y ori-memoryPlatform CLI
codex mcp add ori-memory -- npx -y ori-memoryConfig file
{
"mcpServers": {
"ori-memory": {
"command": "npx",
"args": [
"-y",
"ori-memory"
]
}
}
}Config file
{
"servers": {
"ori-memory": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"ori-memory"
]
}
}
}Platform CLI
gemini mcp add ori-memory npx -y ori-memoryStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about ori-memory
How do I install ori-memory in Claude Code?
Run claude mcp add ori-memory -- npx -y ori-memory to add the ori-memory MCP server to Claude Code.
What does ori-memory do?
It gives an agent persistent memory based on Markdown files and a knowledge graph. Search runs against an index derived from that content.
Where does ori-memory keep its memory?
The source information lives in Markdown files, wiki-links form graph connections, and Git can track versions. Export learned retrieval state before deleting data.
