Skip to main content

LittleHive local-first multi-agent assistant foundation

Project description

LittleHive 🐝

A private, local-first AI executive assistant built exclusively for Apple Silicon.

LittleHive runs entirely on your Mac — powered by Mistral's Ministral models via Apple's MLX framework. Your emails, calendar, and personal data never leave your machine.

No cloud AI. No subscription. Just a fast, intelligent assistant on your hardware.


Features

  • 100% Local AI — Runs Ministral 3B / 8B / 14B natively on Apple Silicon via MLX with KV prompt caching for sub-second responses.
  • Google Workspace — Connects to Gmail, Google Calendar, and Google Tasks. Read, draft, send emails (with PDF attachments); create events; manage tasks.
  • GitHub Integration — Create, list, update, and close GitHub issues. Add comments to issues and PRs directly from chat.
  • Web Search — Search the web via DuckDuckGo for current events, prices, or facts the model isn't confident about.
  • Webpage Reader — Fetch and summarize any webpage by URL. Just paste a link and ask for a summary.
  • Custom APIs — Register external APIs (weather, stocks, smart home, RSS feeds) and the agent calls them on demand. Supports JSON and XML/RSS responses. Auto-geocodes location-based APIs.
  • Shell & File Tools — Run governed shell commands and manage files within a sandboxed workspace. Three-tier security (allowed / logged / blocked) with a full audit trail. Disabled by default.
  • Text-to-Speech — Ask the agent to announce something and it speaks aloud on your Mac using the built-in say command.
  • Proactive Scheduling — Background threads poll for new emails, fire reminders on time, and sync your calendar automatically.
  • Long-Term Memory — Remembers important facts across conversations. Nightly extraction saves key details from your chats.
  • Contacts Directory — Manage stakeholders with optional auto-reply drafting for trusted contacts.
  • Finance Tracking — Track bills, due dates, and mark payments as they come in.
  • Telegram Bot — Chat with your assistant from Telegram with typing indicators and chat ID authorization.
  • Web Dashboard — A local web interface with real-time chat, context usage monitoring, dark mode, and full configuration.
  • Self-Updating — Check for and install updates directly from PyPI with a single command.

Requirements

  • Hardware: Apple Silicon Mac (M1, M2, M3, or M4). Intel Macs are not supported.
  • RAM: 8 GB minimum (3B model), 16 GB recommended (8B model), 24+ GB ideal (14B model).
  • Software: macOS with Python 3.11+.

Installation

1. Create a virtual environment:

python3 -m venv littlehive-env
source littlehive-env/bin/activate

2. Install LittleHive:

pip install littlehive

3. Run the setup wizard:

lhive setup

The wizard walks you through identity, Google OAuth, Telegram, model selection, and preferences. Takes about 2 minutes.


CLI Commands

lhive setup          Interactive setup wizard (run this first)
lhive start          Start the agent in the background
lhive stop           Stop the agent
lhive restart        Restart the agent
lhive status         Show agent status and configuration
lhive update         Check for and install updates from PyPI
lhive version        Show current version
lhive auth google    Re-run Google OAuth flow

The first lhive start downloads the AI model (~2-8 GB depending on model size). Subsequent starts are instant.


Dashboard

Once the agent is running, open your browser:

http://localhost:8080

The dashboard includes:

  • Dashboard — Chat with your agent, view unread emails, today's events, reminders, task queue, and pending bills at a glance.
  • Contacts — Manage your contacts directory and auto-reply settings.
  • Memories — View, edit, or delete facts the agent has memorized.
  • Custom APIs — Register and manage external API integrations.
  • System — Background schedulers, data management (flush queues), and shell audit log.
  • Settings — Configure identity, model, Telegram, GitHub, shell tools, and task preferences.
  • Top Bar — Live clock, context usage %, model name, connection status, and theme toggle.

What Can You Ask?

  • Email: "Do I have unread emails?" · "Send a PDF summary about AI trends to Sarah." · "Archive all newsletters."
  • Calendar: "What's on my schedule tomorrow?" · "Block 2 hours for deep work."
  • Reminders: "Remind me about the dentist at 3 PM." · "What reminders do I have?"
  • Web Search: "What's the latest news on AI?" · "Current weather in London."
  • Webpage: "Summarize this page: https://example.com/article"
  • GitHub: "Create an issue for the login bug." · "List open issues on stackcv." · "Close issue #42."
  • Custom APIs: "What's the weather right now?" · "Register an API for stock prices."
  • Finance: "Add a bill for electricity — $240 due March 20." · "Mark the internet bill as paid."
  • Memory: "Remember that my son's name is Vivaan." · "Who is in my family?"
  • Contacts: "Look up Sarah's email." · "Add John as a contact."
  • Shell: "List files in my workspace." · "Run git status."
  • TTS: "Announce: dinner is ready!"

Chat Commands

Type these directly in the chat window or Telegram:

/reset    Wipe context and start a fresh conversation
/context  Show current token usage and context health
/clear    Clear the chat window (UI only, keeps memory)
/help     Show available commands

Optional Integrations

Integration Setup
Google Workspace Run lhive auth google with your credentials.json from Google Cloud Console
Telegram Create a bot via @BotFather, add the token in Settings
GitHub Generate a Personal Access Token with repo scope, add it in Settings
Custom APIs Register via the dashboard or chat: "Register an API called get_weather..."
Shell Tools Enable in Settings, configure workspace folder and command permissions

Data Storage

Everything stays local:

Path Contents
~/.littlehive/config/ Configuration and Google OAuth tokens
~/.littlehive/db/littlehive.db Chat history, memories, reminders, cached emails, custom APIs, shell audit log
~/.littlehive/logs/agent.log Runtime logs for troubleshooting

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

littlehive-0.7.4.tar.gz (89.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

littlehive-0.7.4-py3-none-any.whl (100.6 kB view details)

Uploaded Python 3

File details

Details for the file littlehive-0.7.4.tar.gz.

File metadata

  • Download URL: littlehive-0.7.4.tar.gz
  • Upload date:
  • Size: 89.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for littlehive-0.7.4.tar.gz
Algorithm Hash digest
SHA256 f3b20cc053b47c1e1822c0b7fd43a07c2fd203f9bcc038792b1d8fe0907c4b3a
MD5 f62cb138d2711fb36f71cb8ec6a7da39
BLAKE2b-256 df87debe7b0dbc3b36e2118b902a31c389501c0402c778192f50855c4a3f7d20

See more details on using hashes here.

Provenance

The following attestation bundles were made for littlehive-0.7.4.tar.gz:

Publisher: publish.yml on stackcv/littlehive

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file littlehive-0.7.4-py3-none-any.whl.

File metadata

  • Download URL: littlehive-0.7.4-py3-none-any.whl
  • Upload date:
  • Size: 100.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for littlehive-0.7.4-py3-none-any.whl
Algorithm Hash digest
SHA256 b8cd980c3b0b334ca44c83dbe7c9db486b1925981fea32c700002454eb178364
MD5 e73265999725aa63aa6230fe0405c03f
BLAKE2b-256 b911a0a1ba9fab260db8e960583c06efeae67c93268ba604b402867d4e55e223

See more details on using hashes here.

Provenance

The following attestation bundles were made for littlehive-0.7.4-py3-none-any.whl:

Publisher: publish.yml on stackcv/littlehive

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page