Skip to main content

A lightweight personal AI assistant framework

Project description

Kyber Logo

Your new favorite AI assistant.

Set up in 60 seconds. Chat from Discord, Telegram, WhatsApp, the web dashboard, or the command line.

PyPI Docs License: MIT

Install In 60 Seconds

curl -fsSL https://kyber.chat/install.sh | bash

Installer flow:

  • Detects macOS, Linux, and WSL automatically
  • Installs dependencies and kyber-chat
  • Guides provider + model configuration
  • Writes secure local config and .env
  • Optionally sets up service mode

Why Kyber

Most assistants are fine for simple Q&A, then fall apart when the real work starts. Kyber is built for actual execution: it can run multi-step tasks, keep you updated while it works, and stay responsive to new messages without getting stuck. You get the speed of chat plus the reliability of a real automation runtime.

  • Fast setup One command to install, configure, and run
  • Actually concurrent Long-running tasks do not block new messages
  • Background execution Start complex work and keep chatting
  • Chat anywhere Discord, Telegram, WhatsApp, dashboard, or CLI
  • Tooling built in Shell, file I/O, web search/fetch, cron, notebook, remote tools, and skills
  • Runs anywhere Laptop, VPS, homelab, Raspberry Pi
  • Multi-machine aware Pair machines, run commands remotely, and share notes across your fleet
  • Observable Real-time task status, progress, and cancellation
  • Secure by default Secrets in .env, not raw config JSON

What You Can Do

Build and Ship

  • Read, edit, and refactor code across repositories
  • Run shell commands for build/test/deploy workflows
  • Debug failures with iterative tool use and live status updates

Research and Decide

  • Search and fetch web sources
  • Extract and summarize articles/docs/transcripts
  • Pull structured outputs for reports and planning

Automate Repetitive Work

  • Schedule recurring tasks with cron
  • Send proactive updates into your channels
  • Keep workflows running without babysitting

Work Across Machines

  • Pair your laptop and VPS so one Kyber can reach the other
  • Run remote commands and inspect files on connected peers
  • Share durable notes through the built-in notebook

Operate Confidently

  • Track active and completed tasks in the dashboard
  • Cancel misfired tasks from UI or API
  • Review security scan findings and history

Provider Flexibility

Kyber works with:

  • OpenAI
  • Anthropic
  • OpenRouter
  • DeepSeek
  • Any OpenAI-compatible endpoint

You can select the provider/model during onboarding and change it later from the dashboard.

Real Workflow Examples

Debug a Production Bug

you: "login endpoint fails on invalid passwords, find and fix it"
kyber: reads code -> isolates validation bug -> patches -> runs tests -> reports diff

Research + Action

you: "summarize latest API changes and open PRs that need migration updates"
kyber: fetches sources -> summarizes impact -> edits files -> opens actionable checklist

Proactive Ops

you: "check inboxes every 30 minutes and send urgent items to Discord"
kyber: schedules cron -> runs checks -> posts concise summaries on schedule

Dashboard Highlights

  • Live chat with your running agent
  • Task timeline with progress + status
  • Cancel/redeliver controls
  • Provider/model configuration
  • Cron job management
  • Skills discovery and installation
  • Notebook visibility across paired machines
  • Security scan controls and report views

Dashboard Preview

Web Chat

Kyber Dashboard Web Chat

Providers

Kyber Dashboard Providers

Agent

Kyber Dashboard Agent Settings

Cron Jobs

Kyber Dashboard Cron Jobs

Skills

Kyber Dashboard Skills

Security Center

Kyber Dashboard Security Center

Discord Bot Personas

Kyber Dashboard Discord Bot Personas

What’s New in the Current Architecture

Kyber now supports a host/spoke network model for paired machines. Your laptop can talk to a VPS, a VPS can expose a notebook and selected tools, and the agent can choose whether work should happen locally, remotely, or as shared context. The command surface is broader too, with notebook tools, remote execution primitives, and richer CLI/TUI plumbing alongside the original local shell/file/web toolkit.

Security And Reliability

  • Secrets are stored in .env with locked permissions
  • Task lifecycle is tracked (queued, running, completed, failed, cancelled)
  • Cancellations are propagated back to the originating channel
  • Web tooling includes clearer diagnostics for DNS/SSL/timeout failures

Project Links

Contributing

Issues, PRs, and feature proposals are welcome. If you are unsure where to start, open an issue with:

  • Use case
  • Desired behavior
  • Current behavior
  • Logs/screenshots

License

MIT

Project details


Release history Release notifications | RSS feed

Download files

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

Source Distribution

kyber_chat-2026.4.21.72.tar.gz (928.5 kB view details)

Uploaded Source

Built Distribution

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

kyber_chat-2026.4.21.72-py3-none-any.whl (945.9 kB view details)

Uploaded Python 3

File details

Details for the file kyber_chat-2026.4.21.72.tar.gz.

File metadata

  • Download URL: kyber_chat-2026.4.21.72.tar.gz
  • Upload date:
  • Size: 928.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.11

File hashes

Hashes for kyber_chat-2026.4.21.72.tar.gz
Algorithm Hash digest
SHA256 6c3a5e1ea21e8ecf9b6172a4f3a28b55267a9455241d7f654717c597f3e2eea8
MD5 1d7d0f9aa0e9ca4ac951f2e783cdc43d
BLAKE2b-256 c19e7f5955e964c51f963a97ff9adb43e208e1ad2dff832f246b1b5619b7b385

See more details on using hashes here.

File details

Details for the file kyber_chat-2026.4.21.72-py3-none-any.whl.

File metadata

File hashes

Hashes for kyber_chat-2026.4.21.72-py3-none-any.whl
Algorithm Hash digest
SHA256 7d1da593cd7270bfd139d462eb961f437160f0e60ce4a6081fd3a502b793e0cb
MD5 811899bd41526f5223086796921f496a
BLAKE2b-256 14958f7a05cf927ab7748eaf9cfe690cae01ab09e1a794ebd1af841de1d53490

See more details on using hashes here.

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