Geniuz

MCP文档知识

提供方 jackccrawford

文档知识 未标注 API Key 社区 / 第三方 Claude DesktopClaudeCursorCodex

Open-source MCP server that gives Claude persistent memory. Your AI keeps your standards, decisions, and reasons across sessions. Recalled by meaning, not keywords. Runs fully local: on-device semantic search, no cloud, no API keys. macOS, Windows, Linux. MIT.

一句话选型结论

按资料字段整理

Geniuz 属于文档知识类 MCP 服务器,传输方式未标注,API Key,可直接接入 Claude Desktop、Claude、Cursor 等,没有现成安装命令,需按官方文档接入。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类文档知识 传输方式未标注 鉴权API Key 部署方式未标注
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

文档知识
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

mcp serverai memorymodel context protocol

安装与配置

未标注 API Key
上游未给出传输方式
  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
文档检索问答

在你的文档库里按问题找答案。

知识库读写

把新资料写进知识库,随用随取。

内容整理

把零散资料归纳成结构化笔记。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 已适配 Claude Desktop、Claude、Cursor,换客户端不用重新研究配置。

先想清楚的地方

  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 使用前要准备好API Key对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在文档知识场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Geniuz当前 未标注 未标注 未标注 无 未标注
dify 未标注 stdio · 本地进程 本地部署 有 2026-04-11
open-webui 未标注 stdio · 本地进程 本地部署 有 2026-04-11
JeecgBoot 未标注 stdio · 本地进程 本地部署 有 2026-04-11
Cherry Studio 未标注 未标注 未标注 无 2026-05-17
mempalace 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

文档知识
字段来自项目仓库与公开索引,中文说明由本站标注
提供方jackccrawford上游标注的维护方 jackccrawford
分类文档知识本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式API Key远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · Cross-platform本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

20.0 KB
来自项目仓库与上游目录的原文,未做删改

Geniuz

Persistent memory for AI agents. Runs on your machine, MIT-licensed, free.

geniuz.life ·
Download ·
Built by mVara


Your AI is sharp. It also starts every conversation from zero. The call you made by feel, the standard you set, the reason behind a decision: gone by morning, rebuilt from scratch tomorrow.

Geniuz fixes that. One install, local, private, searchable by meaning. The context compounds instead of resetting at every boundary. Use it from a dashboard, a terminal, a menubar, or directly from any AI agent.

> New in 2.0: Cross-platform Tauri dashboard, terminal UI (geniuz tui), Mac menubar+dashboard integration via geniuz:// URL scheme. See release notes →
>
> Note: Screenshots in this README are from a pre-2.0.5 build and will refresh with the next release. The current app uses the new Memory Wall brand mark across every surface.

Three surfaces, one memory

Geniuz is one Rust core with several ways to reach it. Pick what fits the moment.

Dashboard

Browse memories, save new ones, search by meaning, see recent threads. Native vibrancy on macOS, system tray on Windows + Linux.

Surfaces: Memories · Remember · Find · Detail · Status · Data & Export · Settings.

Terminal UI · geniuz tui

Same data, terminal-native. For developers and agents who live in a shell. Two-field compose for /remember, semantic search via /find, sort toggle via /reorder.```
$ geniuz tui


Drill into any memory with `/detail ` (or `/random` to surface one at random):

Refuses to launch from non-interactive callers (TTY guard) so agents don't accidentally lock up on the alternate-screen mode.

### Menubar (macOS) / System tray (Windows + Linux)

Ambient presence. Memory count, recent activity, one-click to the dashboard. The menubar app is dock-less on Mac (`LSUIElement`). It's a residence, not a window.

### CLI for agents

Underneath everything, the `geniuz` command. Your agents call it from any shell, any framework.```bash
geniuz remember -c "OAuth token refresh is async but middleware assumed sync. Swapped lines 42-47." -g "fix: auth token refresh — async ordering"
geniuz recall "authentication middleware"

Searched "authentication middleware," found a memory about "OAuth refresh" and "middleware ordering." The meaning matched. No re-investigation. No human re-explaining.

The full surface:```bash
% geniuz --help
Start here: 'geniuz recent' to see what's in your folder.

GENIUZ: Your AI remembers now.

Persistent memory for AI agents. Three R's: remember, recall, recent.
Works with any agent framework — Claude Code, Cursor, Windsurf, Aider,
or anything that can run a shell command.

Usage: geniuz <COMMAND>

Commands:
remember Save a memory — what you learned, decided, or discovered
recall Search your memories — semantic by default
recent Show recent memories
capture Capture files or directories into your folder
watch Watch for new memories in real time
backfill Build embedding cache for semantic search
skill Show usage guide for agents
status Show folder stats
tui Launch the terminal UI for browsing your memories
dashboard Launch the graphical dashboard app
mcp MCP server for Claude Desktop — run, install, or check status
settings User settings — read or change preferences
help Print this message or the help of the given subcommand(s)

Options:
-v, --version
Show version information

-h, --help
Print help (see a summary with '-h')

Examples:
geniuz tui Browse memories in a terminal UI
geniuz remember -c "Fixed the auth bug" -g "fix: token refresh"
geniuz recall "auth" Semantic search
geniuz recent Latest memories
geniuz capture ./notes/ Bulk-load markdown files
geniuz backfill Build embedding cache

Folder: Defaults to .geniuz in your home directory
Override with GENIUZ_HOME to change the folder location
Override with GENIUZ_STATION for a specific memory.db file
Multiple agents can share a folder for shared memory.

Use "geniuz [command] --help" for more information.


---

## Geniuz Free vs Geniuz Team

This repository is **Geniuz Free**: MIT-licensed, single-station memory for any AI chassis. Install it, bring your own AI, get memory that survives across conversations.

**Geniuz Team** is the paid tier from mVara: unlimited continuity, perpetual agents that live in your environment, and a team chassis that coordinates around your customers&#x27; reality. Geniuz Team is built on the same substrate as Geniuz Free.

If you want memory for yourself or a developer team using existing AI tools, Geniuz Free is what you want.

If you want a configured team of agents living in your private cloud, remembering your work, and coordinating around your customers, that&#x27;s Geniuz Team. Contact mVara for pilot inquiries.

## Why local

- **Private.** Your data never leaves your machine. No cloud. No account. No telemetry.
- **Fast.** No network calls. Semantic search runs locally in the binary.
- **Free.** No API keys. No token costs for memory. No subscription.
- **Portable.** Your folder is a SQLite file. Copy it, back it up, share it.
- **Framework-independent.** Switch from Cursor to Claude Code. Your memory comes with you.

---

## Install

Pick the path that matches your setup.

### macOS · one click

Download **Geniuz.dmg**, double-click, drag to Applications. Signed and notarized by Managed Ventures LLC. No Gatekeeper warnings.

One DMG installs three things: the **menubar app** (always-on), the **dashboard** (launched from the menubar&#x27;s &quot;Open Dashboard&quot; or via `geniuz://`), and the **CLI** (bundled at `Geniuz.app/Contents/Resources/geniuz`).

Wire the CLI into your shell PATH if you want it on the command line:```bash
sudo ln -sf /Applications/Geniuz.app/Contents/Resources/geniuz /usr/local/bin/geniuz

Or skip the DMG entirely and use the CLI-first install below. That path installs to ~/.geniuz/bin/ and adds itself to your PATH without sudo. Apple Silicon native; Intel Macs run via Rosetta 2 (universal binary coming).

Windows · one click

Download Geniuz-Setup.exe Windows 11, per-user, no admin prompt. Signed via Azure Trusted Signing (Microsoft-rooted cert, no "unknown publisher" warning). For managed deployment it runs silently with /VERYSILENT /SUPPRESSMSGBOXES /NORESTART.

After install, the dashboard runs as a system tray app. Left-click → window; right-click → menu.

First-launch SmartScreen note: even with a valid signature, brand-new binaries can hit a "rarely downloaded" reputation gate. If you see "Windows protected your PC," click "More info" → "Run anyway." That's per-binary reputation, separate from cert trust.

Mac / Linux · one command (developer path)```bash

curl -fsSL https://raw.githubusercontent.com/jackccrawford/geniuz/main/install.sh | bash


Detects your OS and architecture, downloads the matching CLI binary (with TUI built in), installs to `~/.geniuz/bin/`. Best for developers, fleet operators, and anyone using Claude Code, Cursor, Windsurf, Aider, or any agent framework that can run a shell command.

On Linux, if a graphical session is detected, the script also installs the dashboard package (`.deb` / `.rpm` / `.AppImage`, picked from your package manager) with one `sudo` prompt. Headless servers skip the dashboard step automatically. Opt out explicitly with `GENIUZ_NO_DASHBOARD=1`. On macOS, the script installs only the CLI. Use the DMG above for the dashboard + menubar.

### Linux platform notes

Supported architectures:

- **x86_64** (Ubuntu, Debian, Fedora, Arch; modern distros with glibc 2.34+)
- **arm64** (Raspberry Pi 5, Pi OS / Debian Bookworm+, NVIDIA Jetson, Ampere, AWS Graviton, Oracle Ampere)

The arm64 build bundles ONNX Runtime 1.22 as a sibling `.so` and wraps the CLI with an `LD_LIBRARY_PATH` script, so it runs cleanly on older-glibc systems like Pi OS Bookworm (glibc 2.36). The x86_64 build is a single static binary.

Claude Desktop isn&#x27;t available on Linux, but `geniuz mcp serve` works as a stdio MCP server for any Linux-compatible MCP client (Claude Code, Cursor, Windsurf, Aider, custom agents). The dashboard ships as a `.deb` and `.AppImage` for desktop Linux; the TUI runs in any terminal.

### From source```bash
git clone https://github.com/jackccrawford/geniuz &amp;&amp; cd geniuz
cargo build --release --bin geniuz
cp target/release/geniuz ~/.local/bin/

To also build the dashboard locally:```bash
cd desktop/dashboard
cargo tauri build


(Requires `tauri-cli`. Mac: nothing else. Linux: `libwebkit2gtk-4.1-dev build-essential libxdo-dev libssl-dev libayatana-appindicator3-dev librsvg2-dev`. Windows: WebView2 runtime, typically preinstalled on Win 10/11.)

### Then choose your path

| You use... | Next step |
|------------|-----------|
| Claude Desktop | `geniuz mcp install` → restart Claude Desktop |
| Claude Code / Cursor / Windsurf | Add two lines to your agent&#x27;s instructions (see below) |
| Custom agents | Call `geniuz remember` and `geniuz recall` from any shell |
| Just want to see your memories | `geniuz tui` in any terminal, or open the dashboard from the menubar/tray |

---

## How it works

Geniuz is a compiled Rust binary backed by SQLite. No cloud. No API key. No account. Your data stays on your machine.

- **Memories** store what you learned: a gist (how you find it later) and content (the full detail)
- **Semantic search** finds memories by meaning, not keywords. Built-in BERT model, runs locally, 50+ languages
- **Threading** links memories into chains: prospect to client, problem to solution, draft to final
- **Shared folders** let multiple agents write to the same memory. What one learns, all find```
   Dashboard ─┐
        TUI  ─┼─→ db::DatabaseManager ──→ memory.db (SQLite)
   Menubar  ─┤                                ↑
        CLI ─┘                                │
                                              ↓
                                       ONNX (BERT)

The model downloads once (~118MB) on first search. Every memory after that is embedded automatically. No setup. No configuration.


Works with everything

Platform How
Claude Desktop geniuz mcp install — automatic remember/recall/recent tools
Claude Code Remember from hooks or inline via Bash; or use the TUI for browsing
Cursor / Windsurf / Aider Any agent that can run a shell command
OpenClaw geniuz capture --openclaw imports your existing memory
Custom agents If your agent can exec, it can remember
Just you Dashboard for browse/search/compose; TUI for the same in a terminal

What it looks like

Save what you learned (CLI):```
$ geniuz remember -c "Maria prefers retention over acquisition in Q2. Budget is $40K." -g "client: Maria — Q2 retention focus, $40K budget"
✅ Remembered 7A3B29F1


**Find it later, by meaning:**```
$ geniuz recall &quot;Maria&#x27;s budget priorities&quot;
7A3B29F1 | 2026-03-05 14:23 | client: Maria — Q2 retention focus, $40K budget (0.52)

Browse interactively:```
$ geniuz tui


Opens a terminal UI with `/recent`, `/remember`, `/find`, `/reorder`, `/random`, `/detail`, `/help`. Press `/` to start typing a command; `↑`/`↓` to navigate; `Enter` to open detail; `Ctrl-C` to exit.

**Thread a follow-up:**```
$ geniuz remember -c &quot;Maria approved the retention plan. Starting in April.&quot; -g &quot;client: Maria — plan approved&quot; -p 7A3B29F1
✅ Remembered E5F6A7B8

The full client history, from first meeting to approval, is one chain. Any future session finds the whole story.


Capture existing knowledge

Already have notes, docs, or agent memory files?```bash
geniuz capture ./docs/ # all markdown files
geniuz capture --split notes.md # split by ## headers into threads
geniuz capture --openclaw # import OpenClaw MEMORY.md + daily logs
geniuz capture --dry-run ./notes/ # preview without importing
geniuz backfill # embed everything for semantic search


Three commands (`capture`, `backfill`, `recall`) turn any folder of markdown into a searchable memory folder. Local RAG with zero infrastructure.

---

## Commands```bash
# The three R&#x27;s — remember, recall, recent
geniuz remember -c &quot;what happened&quot; -g &quot;category: compressed insight&quot;
geniuz remember -c @notes.md -g &quot;session: review&quot;
echo &quot;content&quot; | geniuz remember -c - -g &quot;piped: from process&quot;
geniuz remember -c &quot;follow-up&quot; -g &quot;update&quot; -p 98672A90

geniuz recall &quot;topic&quot;                         # semantic search
geniuz recall &quot;topic&quot; --since 7d              # only the past week
geniuz recall &quot;reviews by Cubic&quot;              # Capitalized names come first
geniuz recall --grep Cubic --grep PR          # matching lines, every term
geniuz recall --thread 98672A90               # a whole thread, root first
geniuz recall --keyword &quot;exact words&quot;         # keyword fallback
geniuz recall --random                        # discover something
geniuz recall --full &quot;topic&quot;                  # include full content
geniuz recall --json &quot;topic&quot;                  # JSON output

geniuz recent                                 # latest memories
geniuz recent -l 5                            # last 5
geniuz recent --full                          # with content

# Interactive
geniuz tui                                    # terminal UI (requires a TTY)

# Capture and index
geniuz capture ./docs/                        # bulk-load files
geniuz backfill                               # build embedding cache

# Folder
geniuz status                                 # folder stats
geniuz watch                                  # poll for new memories
geniuz watch --exec &quot;echo {uuid} {gist}&quot;      # trigger on new memories

# Claude Desktop
geniuz mcp install                            # add Geniuz to Claude Desktop
geniuz mcp status                             # check if configured
geniuz mcp serve                              # run MCP server (used internally)

Integration

Add two lines to your agent's instructions:```
When you learn something worth keeping:
geniuz remember -c "what you learned" -g "category: compressed insight"

When you need to remember something:
geniuz recall "what you're looking for"


---

## Architecture

Geniuz Free exposes four user surfaces plus the MCP server for agents:

- **Dashboard** (Tauri) — for browse, search, compose, settings. Cross-platform.
- **TUI** (`geniuz tui`) — same for terminal users and agents that prefer ratatui over GUI.
- **Menubar / system tray** — always-on ambient presence. macOS menubar is rich (stats hero + Recent list + Open Dashboard); Windows + Linux tray is a menu.
- **CLI** — for scripting and direct agent integration. Every subcommand supports `--json` for procedural callers.
- **MCP** — `geniuz mcp serve` is the stdio MCP server. Used by Claude Desktop, Claude Code, and any MCP client.

All surfaces go through one library: `db::DatabaseManager`. Same write path, same read path, same invariants. The dashboard&#x27;s &quot;Remember&quot; button and the CLI&#x27;s `geniuz remember` end up at the same SQL `INSERT`.

**The SQLite file (`memory.db`) is not a public interface.** Schema may change without notice. Invariants (memory immutability enforced by triggers, every-memory-has-an-embedding enforced by the library&#x27;s write transaction) hold at the interface boundary, not at the file boundary. If you want programmatic access, go through MCP or the CLI.

There is no HTTP server in Geniuz Free. Procedural software speaks MCP. This keeps the storage layer free to evolve and the invariants centralized in one place.

---

## Repo layout

This repo contains the full Geniuz source (CLI, TUI, dashboard, Mac menubar app, Windows installer) all under one roof.

Path · What&#x27;s there
`src/` · Rust CLI + TUI + embedding + MCP server source
`src/tui.rs` · Terminal UI (ratatui + crossterm)
`schema/` · SQLite schema for the memory database
`skills/` · `SKILL.md` — the embedded skill guide `geniuz skill` prints
`install.sh` · Linux/Mac CLI installer (the `curl \ · bash` target)
`desktop/Geniuz/` · macOS menubar app (SwiftUI) — Xcode project
`desktop/dashboard/` · Tauri cross-platform dashboard — Rust + HTML/CSS/JS
`desktop/ship.sh` · Mac release pipeline (build → inject CLI + dashboard → sign → notarize → DMG)
`installer/windows/` · Inno Setup script + Trusted Signing scripts
`images/` · Brand assets — logo, icons, social preview
`Cargo.toml` · Rust crate manifest — pinned dependencies, version

Built artifacts are attached to each GitHub release:

- `Geniuz.dmg` — Mac (arm64, Sonoma 14+) — menubar + dashboard + CLI
- `Geniuz-Setup.exe` — Windows installer (Inno Setup, x86_64, Windows 11) — CLI, MCP server and the dashboard with system tray.
- `geniuz-linux-amd64.tar.gz` — Linux x86_64 CLI + TUI
- `geniuz-linux-arm64.tar.gz` — Linux arm64 CLI + TUI (Pi 5 compatible)
- `Geniuz__amd64.deb` / `Geniuz--1.x86_64.rpm` / `Geniuz__amd64.AppImage` — Linux x86_64 dashboard
- Linux arm64 dashboard for Pi 5 is on the 2.0.5 queue (pending CI extension)

---

## For agents

_This section is for you, the one reading this README._

Your next session starts cold. No memory of what you built, what broke, what you decided. Every memory you write is a gift to that future session.

A memory that says &quot;fixed auth bug&quot; saves zero time. A memory that says what broke, why, and how you fixed it saves an hour. Write for the version of you that knows nothing.

Memories compound. A single memory is a note. A folder of memories is institutional knowledge.

### When to remember

- **After solving something hard.** Your next session will hit the same problem.
- **Before your session ends.** Everything in your context window dies with the session.
- **When you discover something non-obvious.** The small facts save the most time.

### When to recall

- **Starting a new task.** You may have already solved part of it last week.
- **When stuck.** The answer might be in a memory from three sessions ago.
- **After compaction.** Your context was compressed. Your folder wasn&#x27;t.

### Meaning finds, exact recalls

Search by meaning to discover; search exactly to recall. Every answer ends
with `names seen:` and `pointers:` (the next hops) and a `searched:` line
that says what was searched, so an empty answer carries its scope. When a
question has a time in it, add `--since 7d`. When you know the name, use
`--grep Name`. When you&#x27;ve found one piece of a conversation, read the rest
with `--thread ID`.

### When to use the TUI

`geniuz tui` only launches with an interactive terminal. Agents calling from a subprocess get a clean refusal with exit code 2. If you want browsing in code, use `geniuz recent`, `geniuz recall`, or the MCP tools. They&#x27;re shell-friendly.

---

## Performance: geniuz-embed

`geniuz-embed` is an optional background process that holds the ONNX model in memory. Every memory after the first is instant.```bash
geniuz-embed &amp;                 # start (auto-exits after 5 min idle)
geniuz remember -c &quot;first&quot;     # 1.6s (model loads)
geniuz remember -c &quot;second&quot;    # 0.04s (model warm)
Mac (Apple Silicon) Raspberry Pi 5
Without embed server 712ms 1,580ms
With embed server 109ms 40ms
Speedup 6.5x 39.5x

Privacy

Your data stays on your computer. Geniuz stores memories in a local SQLite database. The semantic search model runs locally via ONNX Runtime. No data is sent anywhere. The source code is open. Read every line.


Built with

  • Rust — CLI, TUI, MCP server, dashboard backend
  • ratatui + crossterm — terminal UI
  • Tauri 2 — cross-platform dashboard (window-vibrancy on macOS, system tray on Windows + Linux)
  • ONNX Runtime — local semantic search
  • SwiftUI — macOS menubar app
  • MCP — Claude Desktop integration
  • Azure Trusted Signing — Windows code signing

Learn more

  • Website: geniuz.life
  • Geniuz Team (paid tier): mvara-ai
  • Built by: Jack C. Crawford
  • Discussions: GitHub Discussions
  • Issues: GitHub Issues
  • License: MIT

License

MIT

常见问题

7 问
答案由本站依据条目字段整理,不引用上游原文
Geniuz 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

Geniuz 怎么在 Claude Desktop、Claude、Cursor 里配置?

它的传输方式未标注,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Geniuz 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Geniuz 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

Geniuz 支持哪些客户端?

资料标注支持 Claude Desktop、Claude、Cursor、Codex、Zed;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Geniuz 能提供哪些能力?

资料显示能力标签包括 mcp server、ai memory、model context protocol、claude、anthropic、persistent memory。具体工具清单以官方文档为准。

Geniuz 需要什么运行环境?

运行时为 Cross-platform。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 未标注。