@aitytech/agentkits-memory

AgentKits Memory

by AityTech

Persistent Memory System for AI Coding Assistants via MCP

Fast. Local. Zero external dependencies.

Store decisions, patterns, errors, and context that persists across sessions.
No cloud. No API keys. No setup. Just works.

Quick Start • Web Viewer • Features • Ecosystem • agentkits.net

Features

Feature	Benefit
100% Local	All data stays on your machine. No cloud, no API keys, no accounts
Blazing Fast	Native SQLite (better-sqlite3) = instant queries, zero latency
Zero Config	Works out of the box. No database setup required
Cross-Platform	Windows, macOS, Linux - same code, same speed
MCP Server	memory_save, memory_search, memory_recall, memory_list, memory_status
Web Viewer	Browser UI to view, add, edit, delete memories
Vector Search	Optional HNSW semantic similarity (no external service)
Auto-Capture	Hooks for session context, tool usage, summaries

Web Viewer

View and manage your memories through a modern web interface.

npx agentkits-memory-web

Then open http://localhost:1905 in your browser.

Memory List

Browse all stored memories with search and namespace filtering.

Add Memory

Create new memories with key, namespace, type, content, and tags.

Memory Details

View full memory details with edit and delete options.

Quick Start

1. Install

npm install @aitytech/agentkits-memory

2. Configure MCP Server

Add to your .mcp.json (or .claude/.mcp.json):

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["agentkits-memory-server"]
    }
  }
}

3. Use Memory Tools

Once configured, your AI assistant can use these tools:

Tool	Description
memory_save	Save decisions, patterns, errors, or context
memory_search	Search memories using semantic similarity
memory_recall	Recall everything about a specific topic
memory_list	List recent memories
memory_status	Check memory system status

CLI Commands

# Start MCP server
npx agentkits-memory-server

# Start web viewer (port 1905)
npx agentkits-memory-web

# View stored memories (terminal)
npx agentkits-memory-viewer

# Save memory from CLI
npx agentkits-memory-save "Use JWT with refresh tokens" --category pattern --tags auth,security

# Setup hooks for auto-capture
npx agentkits-memory-setup

Programmatic Usage

import { ProjectMemoryService } from '@aitytech/agentkits-memory';

const memory = new ProjectMemoryService({
  baseDir: '.claude/memory',
  dbFilename: 'memory.db',
});
await memory.initialize();

// Store a memory
await memory.storeEntry({
  key: 'auth-pattern',
  content: 'Use JWT with refresh tokens for authentication',
  namespace: 'patterns',
  tags: ['auth', 'security'],
});

// Query memories
const results = await memory.query({
  type: 'hybrid',
  namespace: 'patterns',
  content: 'authentication',
  limit: 10,
});

// Get by key
const entry = await memory.getByKey('patterns', 'auth-pattern');

Auto-Capture Hooks

The package includes hooks for automatically capturing AI coding sessions:

Hook	Trigger	Action
context	Session Start	Injects previous session context
session-init	First User Prompt	Initializes session record
observation	After Tool Use	Captures tool usage
summarize	Session End	Generates session summary

Setup hooks:

npx agentkits-memory-setup

Or manually copy hooks.json to your project:

cp node_modules/@aitytech/agentkits-memory/hooks.json .claude/hooks.json

Memory Categories

Category	Use Case
decision	Architecture decisions, ADRs
pattern	Reusable code patterns
error	Error solutions and fixes
context	Project context and facts
observation	Session observations

Storage

Memories are stored in .claude/memory/memory.db within your project directory.

.claude/memory/
├── memory.db          # SQLite database
└── memory.db-wal      # Write-ahead log (temp)

CJK Language Support

AgentKits Memory has automatic CJK support for Chinese, Japanese, and Korean text search.

Zero Configuration

When better-sqlite3 is installed (default), CJK search works automatically:

import { ProjectMemoryService } from '@aitytech/agentkits-memory';

const memory = new ProjectMemoryService('.claude/memory');
await memory.initialize();

// Store CJK content
await memory.storeEntry({
  key: 'auth-pattern',
  content: '認証機能の実装パターン - JWT with refresh tokens',
  namespace: 'patterns',
});

// Search in Japanese, Chinese, or Korean - it just works!
const results = await memory.query({
  type: 'hybrid',
  content: '認証機能',
});

How It Works

• Native SQLite: Uses better-sqlite3 for maximum performance
• Trigram tokenizer: FTS5 with trigram creates 3-character sequences for CJK matching
• Smart fallback: Short CJK queries (< 3 chars) automatically use LIKE search
• BM25 ranking: Relevance scoring for search results

Advanced: Japanese Word Segmentation

For advanced Japanese with proper word segmentation, optionally use lindera:

import { createJapaneseOptimizedBackend } from '@aitytech/agentkits-memory';

const backend = createJapaneseOptimizedBackend({
  databasePath: '.claude/memory/memory.db',
  linderaPath: './path/to/liblindera_sqlite.dylib',
});

Requires lindera-sqlite build.

API Reference

ProjectMemoryService

interface ProjectMemoryConfig {
  baseDir: string;              // Default: '.claude/memory'
  dbFilename: string;           // Default: 'memory.db'
  enableVectorIndex: boolean;   // Default: false
  dimensions: number;           // Default: 384
  embeddingGenerator?: EmbeddingGenerator;
  cacheEnabled: boolean;        // Default: true
  cacheSize: number;            // Default: 1000
  cacheTtl: number;             // Default: 300000 (5 min)
}

Methods

Method	Description
initialize()	Initialize the memory service
shutdown()	Shutdown and persist changes
storeEntry(input)	Store a memory entry
get(id)	Get entry by ID
getByKey(namespace, key)	Get entry by namespace and key
update(id, update)	Update an entry
delete(id)	Delete an entry
query(query)	Query entries with filters
semanticSearch(content, k)	Semantic similarity search
count(namespace?)	Count entries
listNamespaces()	List all namespaces
getStats()	Get statistics

Requirements

• Node.js LTS: 18.x, 20.x, or 22.x (recommended)
• MCP-compatible AI coding assistant

Node.js Version Notes

This package uses better-sqlite3 which requires native binaries. Prebuilt binaries are available for LTS versions only.

Node Version	Status	Notes
18.x LTS	✅ Works	Prebuilt binaries
20.x LTS	✅ Works	Prebuilt binaries
22.x LTS	✅ Works	Prebuilt binaries
19.x, 21.x, 23.x	⚠️ Requires build tools	No prebuilt binaries

Using Non-LTS Versions (Windows)

If you must use a non-LTS version (19, 21, 23), install build tools first:

Option 1: Visual Studio Build Tools

# Download and install from:
# https://visualstudio.microsoft.com/visual-cpp-build-tools/
# Select "Desktop development with C++" workload

Option 2: windows-build-tools (npm)

npm install --global windows-build-tools

Option 3: Chocolatey

choco install visualstudio2022-workload-vctools

See node-gyp Windows guide for more details.

AgentKits Ecosystem

AgentKits Memory is part of the AgentKits ecosystem by AityTech - tools that make AI coding assistants smarter.

Product	Description	Link
AgentKits Engineer	28 specialized agents, 100+ skills, enterprise patterns	GitHub
AgentKits Marketing	AI-powered marketing content generation	GitHub
AgentKits Memory	Persistent memory for AI assistants (this package)	npm

Star History

License

MIT

Give your AI assistant memory that persists.

AgentKits Memory by AityTech

Star this repo if it helps your AI remember.