Rutter Direct Memory
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Rutter Direct Memory
Install the Memory Hook and MCP for Codex CLI and Claude Code on Windows, Linux, or macOS. Requires Node.js 22+ and an interactive terminal.
中文操作步骤:使用教程。
Run this command on the computer where your AI client runs:
npx --yes --allow-remote=all https://github.com/NyaRu-Kiss/rutter-direct-memory/releases/download/v0.1.3/rutter-direct-memory-0.1.3.tgzOn Windows PowerShell use npx.cmd if script policy blocks npx.ps1.
--allow-remote=all permits this URL download (required by npm 12) without
changing saved npm settings. The package is distributed through GitHub Releases,
not the npm registry under the bare name rutter-direct-memory.
One prompt to connect
The default API is https://efficiencybyai.com/knowledge/v3, with Service ID
default. No server flags, SSH tunnel, or server password are needed. First
installation asks only for your personal Memory User_Key, with hidden input.
Obtain it from your administrator or create one in the server panel's API Key /
User_Key management page. Enter a complete sk-mem- key, never default, a model
API key, or a panel password.
The installer verifies the authenticated asset catalog, backs up configuration, registers Hook and MCP for both clients, and enables the local upload switch. Restart your AI client and confirm its Hook trust prompt if shown. Model API keys and model configuration remain in your AI client.
Repeat the same command to upgrade. Correct public HTTPS configuration is verified and reused without prompting. If an old configuration points to a local forward or another instance, the installer asks for this server's User_Key before migrating it; credentials from a different endpoint are not sent automatically. Failed or cancelled verification preserves the previous configuration.
Related MCP server: Kilonova MCP
Conversations and upload
New conversations ask you to choose Team, Project, and Agent; Task is optional.
Uploading user text and final assistant replies to project L0 is preselected.
Uncheck it to disable upload, or choose no Memory for that conversation. Tools,
subagents, and reasoning are excluded. MCP supplies three read-only tools:
current_assets, search_memory, and list_knowledge.
Every successful install, reinstall, or upgrade enables the upload switch, including when previously disabled. Runtime connection changes and upload 401/403 responses still stop uploads and clear pending messages. Rerun the installer after correcting the configuration to restore uploads for new conversations. Disabled-period history is not automatically backfilled. Reinstalling an already enabled, unchanged identity preserves pending messages.
Options
Append options to the same npx command:
--reconfigure Replace your Memory User_Key
--client codex Configure only Codex
--client claude Configure only Claude Code
--server https://memory.example.com --reconfigure
Advanced setup for another deployment: API, Service ID, key
--uninstall Remove this integration's Hook and MCP configuration
--help Show usage without changing client configurationRuntime files live in ~/.config/rutter-direct-memory/releases/<content-version>/.
Clearing npx cache does not break Hook or MCP. Uninstall keeps credentials,
conversation selections, and upload state. Remote APIs require HTTPS.
MCP dependencies are bundled; there are no install lifecycle scripts. Native
Windows/macOS end-to-end client flows have not been verified for every version.
Source files are maintained in agents/direct-memory-access/ of the main Rutter
repository. Previous tutorials remain available in this repository.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Private memory for AI agents: save and recall facts across Claude Code, Codex, Cursor and ChatGPT.
Project memory, documentation and controlled secret access for AI coding agents.
Persistent cross-session memory shared by Codex, Claude Code, ChatGPT, and other AI agents.
Shared memory for AI tools: save once, recall word for word from Claude, ChatGPT, Codex or Gemini.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides Claude with a persistent local memory and structured knowledge graph to track project states, tasks, and historical decisions across different chat sessions. It enables users to recall information using keyword relevance, time-travel queries, and dependency analysis for complex project management.MIT
- AlicenseAqualityCmaintenancePersistent knowledge base tools for Claude Code that store project context, decisions, and patterns locally across sessions, eliminating cold starts.846 PyPIMIT
- AlicenseNot gradedqualityBmaintenanceEnables local-first personal knowledge management for Codex by turning conversations into a searchable archive, extracting durable facts, building a knowledge graph, and injecting relevant context into later sessions.MIT
- AlicenseNot gradedqualityBmaintenanceEnables saving bookmarks, notes, and files from inside Claude Code or Codex and later finding them from any project or session just by describing what you remember. All storage, indexing, and natural-language search run locally via SQLite and a bundled embedding model, with no accounts, cloud, or API keys.6 npmMIT