cremind 0.0.1

Cremind — open personal assistant server and CLI

Cremind

Open personal assistant — server, desktop app, and CLI you run yourself.

What is Cremind?

Cremind is a personal AI assistant that runs on your own machine or server. It bundles a multi-profile agent runtime, a tool plane, a web UI, and a CLI into one package. Profiles isolate personas, LLM keys, skills, and conversations, so the same install can host a work assistant, a coding assistant, and a home assistant side by side without bleeding context across them.

Under the hood, Cremind speaks two open protocols: A2A for pluggable agents and MCP for pluggable tools. Bring your own LLM (Anthropic, OpenAI, Groq), plug in any MCP server, and reach the assistant through whichever surface fits — browser, desktop app, terminal, or a chat channel like Telegram.

Features

• Multi-profile agents with isolated personas, skills, browser profiles, and conversation history.
• Pluggable LLMs — Anthropic, OpenAI, and Groq, installed on demand via feature flags.
• Tool plane — built-in file browser, terminal, document RAG, and Playwright browser automation; add any MCP server.
• Document ingestion — drop files in a watched folder; Cremind indexes them for retrieval.
• Storage that scales with you — SQLite by default; switch any service to Postgres, Qdrant, or ChromaDB as a Docker sidecar or external endpoint, from the Setup Wizard.
• Chat channels — Telegram bot and userbot adapters included; channel API for adding more.
• Three surfaces — Web UI at :1515, the Cremind App desktop client (Windows / macOS / Linux), and an cremind CLI for chat, conversations, tools, and admin.
• Cross-platform — runs on Linux, macOS, and Windows; sandboxed Docker mode available for stronger isolation.

Install

Cremind App (desktop)

Download the prebuilt installer for your OS from the Releases page:

OS	File
Windows	Cremind App-Windows-<version>-Setup.exe (NSIS)
macOS	Cremind App-macOS-<version>.dmg
Linux	Cremind App-Linux-<version>.AppImage

The desktop app auto-updates from GitHub Releases and bundles the agent — no separate server install needed.

One-line install (Linux / macOS)

curl -fsSL https://cremind.io/install.sh | bash

Or fetch the same script straight from GitHub:

curl -fsSL https://raw.githubusercontent.com/cremind-ai/cremind/main/install/install.sh | bash

One-line install (Windows)

iwr -useb https://cremind.io/install.ps1 | iex

Or fetch the same script straight from GitHub:

iwr -useb https://raw.githubusercontent.com/cremind-ai/cremind/main/install/install.ps1 | iex

The installer detects Docker and asks which mode to use:

• Docker (recommended) — agent runs in a sandboxed XFCE container with VNC at :6080. Postgres / Qdrant / ChromaDB activate as sibling containers on demand.
• Native — Python 3.13+ venv at ~/.cremind/venv, SQLite, agent shares your desktop.

Both modes are idempotent — re-running upgrades in place. See install/README.md for the full flag reference and deployment-type options (local / server / custom).

First run

1. Wait for the installer to finish (it pulls images or builds a venv, runs migrations, and starts the server).
2. Your browser opens automatically to http://<host>:1515/#/setup — the Setup Wizard.
3. Pick an LLM provider, paste an API key, create your first profile, and you're chatting.

Prefer the terminal? cremind chat gives you a streamed REPL against the same profile. cremind --help lists every command (profiles, conversations, tools, agents, channels, LLM config, …).

Requirements

• Docker mode: Docker Engine with Compose v2. Any OS.
• Native mode: Python 3.13.9+ on Linux, macOS, or Windows.

Documentation

• install/README.md — installer reference, flags, deployment types, and file layout.
• CONTRIBUTING.md — development setup and contribution workflow.
• RELEASING.md — release channels (production / test / dev) and the release process.
• In-app docs live under documents/ and are auto-indexed for retrieval.

License

MIT.