mem-zero

mem-zero

Docker 应用程序 from sworcery's 2nd Repository

概述

Self-hosted memory server for AI coding assistants. Store, search, and manage persistent context across sessions so your tools remember what happened last week without stuffing everything into the context window. Each project gets its own isolated vector collection with automatic fact extraction, deduplication, and semantic search. Works with Claude Code, Cursor, Windsurf, and any MCP-compatible client. Includes a bundled LLM for zero-config operation, or connect to Ollama or OpenAI for better quality. Web dashboard included for browsing and managing memories.

运行时参数

网络用户界面
http://[IP]:[PORT:8765]/
网络
bridge
外壳
bash
特权
false
额外参数
--add-host=host.docker.internal:host-gateway

模板配置

WebUI PortPorttcp

Web dashboard and MCP server HTTP port

目标
8765
默认值
8765
价值
8765
Qdrant PortPorttcp

Qdrant API port (optional, for direct access)

目标
6333
StoragePathrw

Persistent storage for Qdrant data

目标
/mem-zero/storage
默认值
/mnt/user/appdata/mem-zero
价值
/mnt/user/appdata/mem-zero
LLM BackendVariable

LLM backend: bundled (self-contained), ollama, or openai. Auto-detected if not set.

目标
LLM_BACKEND
默认值
bundled
价值
bundled
Ollama URLVariable

Ollama API base URL (only used when backend=ollama)

目标
OLLAMA_BASE_URL
LLM ModelVariable

Ollama language model (only used when backend=ollama)

目标
LLM_MODEL
默认值
qwen2.5:7b
价值
qwen2.5:7b
Embedder ModelVariable

Ollama embedding model (only used when backend=ollama)

目标
EMBEDDER_MODEL
默认值
nomic-embed-text
价值
nomic-embed-text
Embedder DimensionsVariable

Embedding vector dimensions

目标
EMBEDDER_DIMENSIONS
默认值
768
价值
768
OpenAI API KeyVariable

API key for OpenAI-compatible backend (only used when backend=openai)

目标
OPENAI_API_KEY
OpenAI Base URLVariable

Base URL for OpenAI-compatible API (works with Groq, Anthropic, etc.)

目标
OPENAI_BASE_URL
默认值
https://api.openai.com/v1
价值
https://api.openai.com/v1
OpenAI ModelVariable

Chat model for OpenAI-compatible backend

目标
OPENAI_MODEL
默认值
gpt-4o-mini
价值
gpt-4o-mini
OpenAI Embed ModelVariable

Embedding model for OpenAI-compatible backend

目标
OPENAI_EMBED_MODEL
默认值
text-embedding-3-small
价值
text-embedding-3-small
Bundled ThreadsVariable

CPU threads for bundled LLM inference

目标
BUNDLED_THREADS
默认值
4
价值
4
API KeyVariable

API key for MCP and REST endpoints (leave empty to disable). Generate one with: openssl rand -hex 32

目标
API_KEY
Dashboard UsernameVariable

Username for web dashboard login (leave empty to disable auth)

目标
DASHBOARD_USER
Dashboard PasswordVariable

Password for web dashboard login (leave empty to disable auth)

目标
DASHBOARD_PASS
DiagnosticsVariable

Enable performance and accuracy metrics on the dashboard

目标
DIAGNOSTICS_ENABLED
默认值
false
价值
false

详细信息

存储库
ghcr.io/sworcery/mem-zero:latest
最后更新2026-05-31
初见2026-05-04

在Unraid 上运行 mem-zero 。

mem-zero 已被列入Unraid OS 的社区应用程序。探索Unraid ,构建灵活的家庭服务器、NAS 或家庭实验室。