kioie avatar

memex

kioie/memex

Gives Cursor, Claude Desktop, and other MCP clients a SQLite-backed memory that survives across chats. You tell it once to remember your coding preferences, stack decisions, or project conventions, then open a new conversation later and the agent recalls them without re-prompting. Ten tools cover the basics: remember to store a fact, retrieve_context to pull relevant memories sized for your token budget, recall for keyword search, update_memory to revise entries with history preserved. Everything stays in a local database file under ~/.memex. Useful when you're tired of re-explaining your testing style or branch naming rules every session. Optional hybrid search mode helps with paraphrased queries. Built in Go, runs via stdio, requires no cloud account or API keys.

GitHub stars
1
tools
10
transport
STDIO
env vars
3
last updated
May 25, 2026

Tools

10 tools

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

rememberSave a fact

Save a fact

No parameter schema in public metadata yet.

retrieve_contextFind relevant facts, sized for the model context window

Find relevant facts, sized for the model context window

No parameter schema in public metadata yet.

recallSearch by keywords (query required)

Search by keywords (query required)

No parameter schema in public metadata yet.

list_memoriesBrowse stored facts without searching

Browse stored facts without searching

No parameter schema in public metadata yet.

update_memoryChange an existing fact (keeps history)

Change an existing fact (keeps history)

No parameter schema in public metadata yet.

get_memoryLook up one fact by ID

Look up one fact by ID

No parameter schema in public metadata yet.

forgetRemove one fact (soft delete)

Remove one fact (soft delete)

No parameter schema in public metadata yet.

delete_memoriesRemove several facts

Remove several facts

No parameter schema in public metadata yet.

delete_all_memoriesWipe a user's memories (confirm=true)

Wipe a user's memories (confirm=true)

No parameter schema in public metadata yet.

memory_historySee how a fact changed over time

See how a fact changed over time

No parameter schema in public metadata yet.

Sponsors

Configuration

MEMEX_DIR

Data directory (default ~/.memex in native install; use /data in Docker)

MEMEX_USER_ID

Default user scope for memories

MEMEX_HYBRID

Set to 1 to enable local vector retrieval fusion

Readme

memex — long-term memory for AI coding agents

Stop re-explaining your stack every chat. Long-term memory for AI coding agents — stored on your machine.

memex gives Cursor, Claude Desktop, and other MCP clients a memory that survives across chats. Save preferences, decisions, and facts once; the agent finds them again later. No cloud account, no API keys, no vector database to run.

Unit Tests Integration Tests Go Reference

Current release: v0.6.0 · Requires: Go 1.26+


When to use memex

Use memex if you:


View the full README on GitHub

Details

Repository
github.com/kioie/memex
Website
github.com/kioie/memex
Categories
Databases, AI & LLM Tools
Package
ghcr.io/kioie/memex:0.6.0
Transport
STDIO
Registry
MCP registry, active

More from kioie

Related Databases MCP servers

View all
1ysydhc avatarTeamMemoryysydhc/team_memoryTeam experience database - let AI accumulate, retrieve and apply team knowledge across chats.1stars
2annasmazhar avatarPySpark MCPannasmazhar/pyspark_mcpSQL migration, AWS Glue job generation, and PySpark optimization as an MCP server.0stars
3arpe-io avatarMigratorxpress Mcparpe-io/migratorxpress-mcpMCP server for MigratorXpress — cross-platform database migration with parallel transfer0stars
4aryanduntley avatarAifparyanduntley/aifpDatabase-driven FP enforcement and project management for AI-maintained codebases0stars
5lutolabs avatarMyAITwinlutolabs/mytwin-mcpA personal RAG database you build from chat, so AI creates work that sounds like you.0stars
6cyanheads avatarfaostat-mcp-servercyanheads/faostat-mcp-serverUN FAOSTAT global food & agriculture statistics over a local SQLite mirror, via MCP.0stars