Claude MarketplacesSubmit
sachitrafa avatar

YourMemory

sachitrafa/yourmemory

This gives Claude persistent memory that works like human recall: memories decay over time unless reinforced, and related facts surface together through an entity graph. It stores conversational context across sessions using SQLite (or Postgres for teams), applies Ebbinghaus forgetting curves to prune stale information every 24 hours, and automatically merges or replaces outdated facts. The MCP tools let Claude remember, recall, update, and forget information. It includes a web dashboard at localhost:3033 to browse what's stored and see memory strength scores. Setup is one command (yourmemory-setup) that detects your client and writes the config. Works with Claude Desktop, Cursor, Cline, Windsurf, or any stdio MCP client.

GitHub stars
244
tools
3
transport
STDIO
env var
1
last updated
May 2, 2026
View on GitHub

Tools

3 tools

Public tool metadata for what this MCP can expose to an agent.

recall_memoryRecall relevant memories using hybrid BM25 + vector retrieval with entity graph expansion. Returns top-k memories ranked by relevance, with graph-expanded neighbours included.4 params

Recall relevant memories using hybrid BM25 + vector retrieval with entity graph expansion. Returns top-k memories ranked by relevance, with graph-expanded neighbours included.

Parameters* required
current_pathstring
Optional file or directory path for spatial memory boost. Memories tagged with this path score higher.
querystring
The search query to find relevant memories. Use keywords from the current task or conversation.
top_knumber
Number of top memories to return. Defaults to 5.default: 5
user_idstring
Unique identifier for the user whose memories to search.
store_memoryStore a new memory. Automatically deduplicates, embeds, and indexes with BM25 and entity graph edges. Skips exact duplicates and bumps recall count instead.5 params

Store a new memory. Automatically deduplicates, embeds, and indexes with BM25 and entity graph edges. Skips exact duplicates and bumps recall count instead.

Parameters* required
categorystring
Memory category controlling decay half-life. Options: 'fact' (~24 days), 'strategy' (~38 days), 'assumption' (~19 days), 'failure' (~11 days).one of fact · strategy · assumption · failuredefault: fact
contentstring
The memory content to store. Write as a single factual sentence: 'Sachit prefers Python' or 'The project uses DuckDB'.
importancenumber
Importance score from 0.0 to 1.0. Controls decay rate — higher importance = slower decay. Use 0.9 for core facts, 0.7 for preferences, 0.5 for regular facts, 0.2 for transient context.default: 0.7
user_idstring
Unique identifier for the user this memory belongs to.
visibilitystring
Visibility scope. 'shared' means all agents can read it. 'private' means only the storing agent can read it.one of shared · privatedefault: shared
update_memoryUpdate an existing memory by ID. Re-embeds the new content, replaces the old memory, and logs the change to the audit trail.3 params

Update an existing memory by ID. Re-embeds the new content, replaces the old memory, and logs the change to the audit trail.

Parameters* required
importancenumber
Updated importance score from 0.0 to 1.0.default: 0.7
memory_idstring
The unique ID of the memory to update. Obtain this from a prior recall_memory call.
new_contentstring
The updated memory content. Write as a single factual sentence replacing the old content.

Sponsors

Promote your product hereReach 125k+ visitors monthlyPromote your product hereReach 125k+ visitors monthlyAgent CaffeineSponsoredClose your lid, come back to finished workSnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthlyWaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthlyknobsSponsoredSee your site on any phone, any day, in any language

Configuration

PYTHONIOENCODING

Set to utf-8 to ensure correct text encoding on all platforms.

Readme

YourMemory

YourMemory

Your AI has the memory of a goldfish. Not anymore.

Persistent, self-improving memory for AI agents — built on the science of how humans remember.

PyPI PyPI Downloads Python License: CC BY-NC 4.0 GitHub Stars

View the full README on GitHub

Details

Repository
github.com/sachitrafa/yourmemory
Category
AI & LLM Tools
Package
yourmemory
Transport
STDIO
Registry
MCP registry, active

More from sachitrafa

Related AI & LLM Tools MCP servers

View all
1codeofaxel avatarKilncodeofaxel/kilnSpawn an object into existence: describe it, and your agent designs it, checks it, and 3D prints it.242stars
2mpp32 avatarMpp32 Mcp Servermpp32/mpp32Payment layer for AI agents. One MCP, five protocols, thousands of paid APIs your agent can call.242stars
3roomi-fields avatarNotebookLM MCProomi-fields/notebooklm-mcpGoogle Gemini Notebook (ex NotebookLM) over MCP + a local REST API — cited answers, Studio, history240stars
4nikhils4 avatarUI Beatsnikhils4/ui-beatsMCP server for UI Beats. Let your coding agent search, read and install animated React components.235stars
5glifxyz avatarGlifglifxyz/glif-mcp-serverGenerate images, video, and audio with Glif's media-generation agent218stars
6antonpk1 avatarTiny Cryptographyantonpk1/gibber-mcpTiny MCP server with cryptography tools, sufficient to establish end-to-end encryption between LLM agents211stars