← Back to directory

Shodh-Memory MCP Server

Official

A local, LLM-free cognitive memory MCP server.

GitHub source repository ↗

★ 276 Stars Category · Database Very popular Source revision 98c6e4861847

59FMRS · C

Reliability

9/20

Security and permissions

11/20

Maintenance

12/20

Documentation

14/20

Setup experience

13/20

The supplied materials describe an Apache-2.0 licensed, version 0.2.0 MCP server using stdio transport for local cognitive memory. Its main capabilities are LLM-free semantic search and knowledge graphs, together with memory, session, todo, project, reminder, statistics, and backup management.

Read the FMRS scoring method →

Shodh-Memory provides persistent cognitive memory for AI agents and robots. It performs semantic search, Hebbian learning, memory decay, knowledge-graph construction, causal tracing, and related-memory activation with local algorithms, without cloud services, external databases, or LLM calls during storage and recall. The server uses stdio MCP transport and connects to a backend that defaults to http://127.0.0.1:3030.

Tools

remember

Stores a memory.

recall

Searches memories semantically.

recall_by_tags

Searches memories by tags.

proactive_context

Retrieves proactive memory context relevant to the current context.

context_summary

Creates a summary of memory context.

list_memories

Lists memories.

read_memory

Reads a specified memory.

forget

Deletes or forgets a specified memory.

quick_recall

Quickly retrieves memories.

query

Queries memories and related knowledge.

topic

Retrieves memories for a topic.

what_i_know

Queries information known by the system.

recent_memories

Lists recent memories.

pending_work

Queries pending work.

count

Counts memories.

memory_health

Checks memory-system health.

session_summary

Retrieves a session summary.

session_digest

Creates a session digest.

session_history

Queries session history.

fact_narratives

Retrieves fact narratives.

purge_facts

Purges fact records.

add_todo

Adds a todo item.

list_todos

Lists todo items.

update_todo

Updates a todo item.

complete_todo

Completes a todo item.

delete_todo

Deletes a todo item.

reorder_todo

Reorders todo items.

list_subtasks

Lists subtasks.

add_todo_comment

Adds a comment to a todo item.

list_todo_comments

Lists todo comments.

update_todo_comment

Updates a todo comment.

delete_todo_comment

Deletes a todo comment.

todo_stats

Retrieves todo statistics.

add_project

Adds a project.

list_projects

Lists projects.

archive_project

Archives a project.

delete_project

Deletes a project.

set_reminder

Sets a reminder.

list_reminders

Lists reminders.

dismiss_reminder

Dismisses a reminder.

memory_stats

Retrieves memory-system statistics.

verify_index

Verifies the index.

repair_index

Repairs the index.

token_status

Checks token status.

reset_token_session

Resets the token session.

consolidation_report

Retrieves a memory-consolidation report.

backup_create

Creates a backup.

backup_list

Lists backups.

backup_verify

Verifies a backup.

backup_restore

Restores a backup.

backup_purge

Purges backups.

Setup

Ensure the local Shodh-Memory backend is running at the default address, http://127.0.0.1:3030, then add the server with: claude mcp add shodh-memory -- npx -y @shodh/memory-mcp. Claude Desktop or Cursor can use the supplied MCP configuration. For a remote backend, set SHODH_API_URL and, when required, SHODH_API_KEY; SHODH_USER_ID defaults to default.

claude_desktop_config.json

{"mcpServers":{"shodh-memory":{"command":"npx","args":["-y","@shodh/memory-mcp"]}}}

Generate and validate client config

Fit and risk

Best for

AI agents that need offline, low-latency, local memory.
Developers who want to avoid LLM calls during memory storage and recall.
Projects requiring persistent memory, knowledge graphs, or Hebbian learning.
Teams using Zenoh, ROS2, or robotic mission memory.

Not for

Users seeking a hosted cloud memory service or built-in multi-tenant authorization.
Applications expecting the server itself to perform LLM inference.
MCP clients operating without an available local backend.
Workflows that only need simple stateless text processing.

Required permissions

Access to read and write Shodh-Memory's local memory data.
Permission to launch the npm package and local MCP server over stdio.
Network access to the address specified by SHODH_API_URL when using a remote backend.
SHODH_API_KEY when authentication is enabled; this variable is secret.
The tools can create, modify, and delete memories, todos, projects, reminders, and backups.

Risks and side effects

Memories, todos, projects, and backups may contain sensitive information and persist locally.
forget, delete, purge, and backup_restore can cause data loss or overwrite data.
For remote deployments, the deployer must protect the API key and network exposure.
SHODH_USER_ID is a logical memory namespace, not an authorization tenant.
A local model is downloaded on first run, requiring local resources and model storage.

Troubleshooting

Confirm that the Shodh-Memory backend is running at http://127.0.0.1:3030.
Check that SHODH_API_URL points to the correct backend.
For remote backends, verify that SHODH_API_KEY is set and valid.
Confirm that npx is available and that npx -y @shodh/memory-mcp starts successfully.
Use shodh status or shodh doctor to check backend health.
If indexing is abnormal, inspect verify_index, repair_index, and memory_health.

Use cases

Persisting cross-session memory for Claude Code, Claude Desktop, or Cursor.

Running local semantic search, tag search, and knowledge-graph queries.

Managing an agent's todos, projects, reminders, and session summaries.

Storing mission, sensor, and action-outcome memories for robots and edge devices.

Supported clients

Claude CodeFull support

Claude DesktopFull support

CursorFull support

Related MCP servers

Compare with other MCP servers