Skip to main content
Glama

Memory Lite — free local memory for Claude & Cursor (MCP)

A tiny Model Context Protocol server that gives your AI assistant a memory that lives only on your computer (one SQLite file). No cloud, no API keys, zero dependencies (Python 3.9+ standard library). MIT licensed. Made by Jipbap Kitchen.

Tool

What it does

remember

save a fact / preference / decision (optional tags)

recall

keyword match (any word), newest first, max 10 results

forget

delete a memory by id

Data file: ~/.memory-lite/memory.db (override with MEMORY_LITE_DB).

Install

With uv (recommended): nothing to install, the client runs uvx memory-lite-mcp. Or: pip install memory-lite-mcp, or clone this repo and run python -m memory_lite_mcp with PYTHONPATH=src.

Check it works: uvx memory-lite-mcp doctor

Claude Desktop

Edit claude_desktop_config.json

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

{ "mcpServers": { "memory-lite": { "command": "uvx", "args": ["memory-lite-mcp"] } } }

Fully quit and reopen Claude Desktop.

Cursor

~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project) — same JSON as above.

Claude Code

claude mcp add memory-lite -- uvx memory-lite-mcp

Related MCP server: ai-memory

Lite vs. full edition

Memory Lite (free, MIT)

AI Memory Bible (paid)

remember / recall / forget

✅

✅

Search

keyword match, newest first, max 10

SQLite FTS5 BM25 ranking + highlighted snippets

Projects & tag filters

–

✅

update, list, briefing (session-start summary)

–

✅

JSON export / import (backup, move PCs)

–

✅

One-command installers (Windows / macOS / Linux) + auto config with backup

–

✅

doctor self-check + illustrated PDF setup guide & FAQ

basic doctor

✅

Get the full edition: https://bamsuni.com/ai-shop/en/  ·  Payhip: https://payhip.com/b/Jl3Tm

Privacy

Everything stays in a local SQLite file. The server makes no network requests. Don't store passwords or API keys.

License

MIT © 2026 Jipbap Kitchen

Available Tools

3 tools
forgetA
Destructive

Delete a memory by id.

Inputs: id (required).

recallA
Read-only

Find saved memories containing any of the given keywords (newest first, max 10). Use before saying you don't know something about the user.

Inputs: query.

rememberB

Save a fact, preference or decision to local memory.

Inputs: content (required), tags.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 3 tools

    Disambiguation5/5

    The three tools map to unmistakable lifecycle phases: write (remember), read (recall), and delete (forget). There is no overlap in purpose and an agent can pick correctly from the name alone.

    Naming Consistency5/5

    All three names are single, lowercase verbs in a consistent style and follow a coherent memory metaphor (remember/recall/forget). No mixing of conventions or noun/verb forms.

    Tool Count5/5

    Three tools is exactly right for a 'lite' memory server; each one earns its place and nothing peripheral is bolted on. The surface stays small and predictable.

    Completeness3/5

    Create, keyword-read, and delete are covered, but there is no update/edit operation and no way to enumerate stored memories without guessing keywords (recall is capped at 10 and requires terms). Agents must fall back to forget-then-remember to correct a fact, which is a notable but workable gap.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      C
      maintenance
      Persistent semantic memory for AI agents. SQLite-backed, local-first, zero config. Semantic search via Ollama embeddings with keyword fallback. Tools: remember, recall, history, forget, stats.
      17
      37 npm
      1
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Persistent memory for any AI assistant. Zero token cost until recall. Stores memories in local SQLite, ranks by 6-factor scoring, returns results 79% smaller than JSON. Works with Claude, ChatGPT, Grok, Cursor, Windsurf, and any MCP client.
      57
      Apache 2.0
    • A
      license
      Not graded
      quality
      D
      maintenance
      Provides persistent, searchable memory for Claude Code using local SQLite, semantic embeddings, and full-text search, enabling Claude to recall and retrieve context across sessions and projects without external services.
      23 npm
      4
      MIT