mnemosis-mcp
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mnemosis-mcpremember that I prefer Chinese for technical discussions"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Mnemosis
把 AI 的记忆,从“无限存储 + 搜索”改造成“会记住、会遗忘、会整理、会自我怀疑”的系统。
Mnemosis is a human-inspired memory layer for AI agents. Most "AI memory" systems are just storage with semantic search bolted on: they save everything and recall by similarity. Mnemosis instead treats memory as a lifecycle — remembering, reinforcing, consolidating, forgetting, and reconciling — the way human memory actually works.
中文说明:README.zh-CN.md · English: README.md
Install
pip install git+https://github.com/liyexiaoyi/Mnemosis.gitZero runtime dependencies (pure Python stdlib + SQLite). No server, no cloud embeddings required — optional embedder hooks only.
PyPI 版(
pip install mnemosis)发布后会在这里同步更新。
SQLite 存储适合单进程/低并发场景;多个 Agent 并发写入时建议串行访问或接入 外部数据库适配层。
Related MCP server: aimemory
Quick start
from mnemosis import MemoryEngine
from mnemosis.types import MemoryKind, SourceRecord, SourceType
engine = MemoryEngine("memory.db") # pass a path for persistence
engine.remember(
"The user prefers Chinese for technical discussions.",
kind=MemoryKind.SEMANTIC,
source=SourceRecord(origin=SourceType.USER),
cues=["user", "language", "preference"],
importance=0.9,
)
for r in engine.recall("what language does the user prefer?", top_k=3):
print(f"[{r.item.kind.value}] {r.score:.2f} {r.item.content}")
check = engine.check("what is the user's favorite movie?")
print("knowledge gaps:", check.gaps or "none")
engine.sleep() # offline consolidation: dedupe, promote, detect contradictions中文快速开始
pip install git+https://github.com/liyexiaoyi/Mnemosis.gitfrom mnemosis import MemoryEngine
from mnemosis.types import MemoryKind, SourceRecord, SourceType
engine = MemoryEngine("memory.db")
user = SourceRecord(origin=SourceType.USER)
engine.remember(
"用户喜欢用中文讨论技术问题。",
kind=MemoryKind.SEMANTIC,
source=user,
cues=["语言", "偏好"],
importance=0.9,
)
engine.remember(
"昨天一起修了 SQLite 锁死的问题。",
kind=MemoryKind.EPISODIC,
source=user,
cues=["SQLite", "锁死"],
)
for r in engine.recall("用户用什么语言聊天?", top_k=3):
print(f"[{r.item.kind.value}] 相关度 {r.score:.2f} {r.item.content}")一分钟完整演示(记住 → 检索 → 新旧矛盾 → 睡眠整合 → 元认知 → 遗忘回收):
pip install git+https://github.com/liyexiaoyi/Mnemosis.git
python examples/demo.py # 仓库内不想安装?可以在线体验:
Google Colab 打开演示笔记本 (部分地区访问不了 Colab,可改用下面的方式)
GitHub Codespaces 一键打开:云端环境, 打开终端执行
python examples/demo.py即可下载
examples/Mnemosis_demo.ipynb后,用本地 Jupyter 或百度 AI Studio 打开
Use with your AI client (MCP)
One-line MCP integration for Claude Desktop, Cursor, Codex and any MCP client:
{
"mcpServers": {
"mnemosis": {
"command": "mnemosis-mcp",
"args": ["--db", "/path/to/memory.db"]
}
}
}Full guide (including Cursor and Codex configs): docs/mcp-quickstart.md.
Command line
mnemosis --db memory.db remember "用户喜欢用中文讨论技术问题。" --kind semantic
mnemosis --db memory.db recall "用户喜欢什么语言?"
mnemosis --db memory.db sleep
mnemosis --db memory.db check "用户最喜欢的电影是什么?"
mnemosis mcp --db memory.db # or: mnemosis-mcp --db memory.dbDocs & research basis
docs/memory-model.md— the memory modeldocs/research.md— human-memory papers behind each featuredocs/architecture.md— architecturedocs/mcp-quickstart.md— MCP integrationdocs/roadmap.md— roadmapCHANGELOG.md— every round of iteration, with measured results
Testing
python -m unittest discover -s tests -q # 320 unit tests
python benchmarks/locomo_bench.py --mode keyword # LoCoMo-style long dialogueLicense
MIT. See LICENSE.
Contributing
PRs, issues and new benchmark scenarios are welcome — see CONTRIBUTING.md. Every change should come with a test or a measured benchmark result.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Alicense-qualityAmaintenanceAn MCP-native, local-first memory server that gives AI agents persistent, structured memory across sessions and tools, enabling them to maintain identity and context without reconfiguration.3MIT
- Alicense-qualityDmaintenanceGives AI agents persistent memory with semantic search, automatic extraction, and memory decay, accessible via MCP protocol.7MIT
- Alicense-qualityDmaintenanceLocal-first AI memory layer with hybrid retrieval and brain-inspired namespaces. Enables agents to save, search, and manage memories directly via MCP tools.1MIT
- Alicense-qualityBmaintenanceEnables AI agents to have persistent, self-managing memory with bi-temporal supersession, timely forgetting, and recall under a limited context window, using MCP protocol.MIT
Related MCP Connectors
User-owned memory for AI agents, Copilot, Claude, IDEs, CLIs, and chat apps over remote MCP.
Your memory, everywhere AI goes. Build knowledge once, access it via MCP anywhere.
Private-by-default, local-first memory/context/task orchestrator for MCP apps and agents.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/liyexiaoyi/Mnemosis'
If you have feedback or need assistance with the MCP directory API, please join our Discord server