OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

LINGSHU

LINGSHU is an open-source memory system for AI agents that records knowledge in a plain-Markdown cognitive graph and makes it available for retrieval across sessions. It provides a standard-input/output MCP server and a DSH plugin integration, with support described for sharing memory among multiple agents.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

LINGSHU — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use LINGSHU when building or configuring an agent that needs readable, persistent memory across sessions, especially when MCP connectivity or shared memory for multiple agents is relevant.

Who it is for

  • Developers adding persistent memory to an AI agent.
  • Teams connecting MCP-capable agents to a shared memory store.
  • DSH users integrating memory through the documented plugin.

Documented uses

  • Preserve information between agent sessions.
  • Retrieve stored facts or agreements for a later conversation.
  • Inspect and edit agent memory as Markdown.
  • Connect memory tools to an MCP-compatible agent host.
Before you choose

Requirements & limits

The repository is public and displays an MIT license. The documented setup requires local developer tooling; it is not presented as a hosted end-user service. Memory writes require credentials; without them, the documented guest mode is read-only.

  • The MCP server is stdio-based, so the agent host must support an appropriate MCP connection.
  • The quick-start init command generates configuration only; it does not launch the service or create the memory store.
  • Write credentials are needed for durable memory writes; guest mode supports reading and recall but not writing.
  • The DSH plugin path specifies Node.js 22.19 or later and a Python interpreter; the bundled memory engine has no third-party Python runtime dependencies.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Generate an MCP configuration snippet for an agent host

Create the LINGSHU configuration snippet for a supported MCP host.

Before you start

  • Node.js 22.19 or later.
  • A Python interpreter; the default is selected by platform, or specify one when configuring.
  • Choose a supported host and memory root directory.

Steps

  1. Run the command below.

    Documented command
    npx @furongjun1999/dsh-memory init

  2. Answer the prompts for host, memory root, and Python interpreter; alternatively run the command below.

    Documented command
    npx @furongjun1999/dsh-memory init -- --end claude --root /path/to/memory --python python3

  3. Use the printed MCP configuration or the generated lingshu-mcp-snippet.json; place the snippet in the host configuration location described by the project, then follow its guidance for credentials and self-checks.

Expected outcome

The initializer produces an MCP configuration snippet and follow-up instructions; it does not create the memory directory, start a service, or write to the memory store.

Guide limitations

  • This is configuration generation only, not a service startup or memory-write procedure.
  • Without write credentials, LINGSHU runs as read-only guest; writes do not persist.
  • Host configuration locations differ by host; consult the linked README for the host-specific steps.
Documentation 1 ↗
Source-based practical guide

Build and install the LINGSHU DSH plugin

Build the plugin and add it to a DSH profile.

Before you start

  • Node.js 22.19 or later.
  • DSH kernel 0.1.2-rc.1 or later; the README reports verification through DSH 0.17.2.
  • A Python interpreter; md_cg is bundled and requires no pip-installed engine.

Steps

  1. Clone https://github.com/FuRongJun-1999/dsh-memory.git and change into the dsh-memory directory.

  2. Run the command below.

    Documented command
    npm install && npm run build

  3. Install the plugin into the DSH profile with dsh plugin --profile web add ..

  4. Enable it in /cordis.yml, using the example configuration at dsh/cordis.yml.example; restart DSH and check that the agent lists the cg and stg tools.

Expected outcome

The plugin is built and added to the selected DSH profile; after enabling and restarting, its memory tools should be available.

Guide limitations

  • Write permission is off by default: without credentials the plugin operates as read-only guest.
  • Use an absolute path if setting a custom memory root; a package-relative path can place memory inside the plugin directory, where package updates may remove it.
  • If the tools fail to register, the README directs users to check the Python interpreter name and the bridge debug log.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
FuRongJun-1999/dsh-memory · Official homepage title · retained in its source language
How we review sourcesSuggest a correction