MemoryGate
Provides durable memory for AI agents with structured storage, semantic search, OAuth authentication, and lifecycle controls.
README
MemoryGate
MemoryGate is a production-ready MCP server that provides durable memory for AI agents. It combines structured storage with semantic search, OAuth-based authentication, and lifecycle controls for retention and archiving.
Highlights
- 20 MCP tools for storage, retrieval, knowledge graph, and cold storage.
- OAuth 2.0 + PKCE for MCP clients, plus Google/GitHub login for users.
- API keys with bcrypt hashing and prefix-based lookup.
- Postgres + pgvector for semantic search; SQLite fallback for local/dev.
- Retention engine with hot/cold tiers, summaries, tombstones, and archives.
- Metadata-only audit logging.
- Rate limiting, security headers, and request size limits by default.
Architecture
- core/ contains shared business logic and models.
- app/ wires FastAPI middleware, routes, and OAuth flows.
- core/mcp registers FastMCP tools and exposes SSE/streamable HTTP.
Endpoints:
- /mcp/ - SSE MCP transport (authenticated).
- /MemoryGate - Alias endpoint for streamable HTTP MCP.
- /auth/* - OAuth login, client credentials, API key management.
- /oauth/* and /.well-known/* - MCP OAuth discovery and PKCE flow.
- /health - Health check.
Quick Start (Local)
Install deps:
pip install -r requirements.txt
pip install -r requirements-postgres.txt
Configure environment (Postgres + pgvector):
export DATABASE_URL="postgresql://user:password@localhost:5432/memorygate"
export OPENAI_API_KEY="sk-..."
export PSTRYDER_DESKTOP="client-id"
export PSTRYDER_DESKTOP_SECRET="client-secret"
export PSTRYDER_DESKTOP_REDIRECT_URIS="https://claude.ai/api/mcp/callback"
Run:
uvicorn app.main:asgi_app --host 0.0.0.0 --port 8080
Verify:
curl http://localhost:8080/health
For local dev without auth, you can set REQUIRE_MCP_AUTH=false.
Agent Usage (Example)
# Bootstrap
memory_bootstrap(ai_name="Kee", ai_platform="Claude")
# Initialize a session
memory_init_session(
conversation_id="uuid",
title="OAuth integration",
ai_name="Kee",
ai_platform="Claude",
)
# Store an observation
memory_store(
observation="Completed OAuth 2.0 + PKCE integration",
confidence=0.95,
domain="technical_milestone",
evidence=["Deployment successful"]
)
MCP Tool Set
- Session and docs: memory_init_session, memory_bootstrap, memory_user_guide
- Storage: memory_store, memory_store_document, memory_store_concept, memory_update_pattern
- Retrieval: memory_search, memory_recall, memory_get_concept, memory_get_pattern, memory_patterns, memory_stats
- Knowledge graph: memory_add_concept_alias, memory_add_concept_relationship, memory_related_concepts
- Cold storage: search_cold_memory, archive_memory, rehydrate_memory, list_archive_candidates
Documentation
See docs/INDEX.md for the full documentation set.
License
Apache 2.0 - see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。