oguzbilgic/kern-ai

What is it?

Key features

• 01Connects one persistent agent session to Slack, Matrix, Telegram, Discord, IRC, Nostr and a terminal.
• 02Runs long shell commands as background jobs and reports results in the originating chat.
• 03Segments and summarizes conversations, with semantic recall over past messages.
• 04Keeps notes and knowledge files in a git repo the agent maintains.
• 05Spawns parallel read-only sub-agents and supports skills and MCP servers.
• 06Deploys each agent as its own Docker container with a persistent volume.

Use cases

• Running an ops or build-monitoring assistant that posts job results into a Slack channel.
• Giving a team a shared research agent reachable from Telegram or Discord.
• Operating a self-hosted agent on a homelab with local models through Ollama.

Quick start

1. 1

   Start an agent on Slack with Docker, passing your model key and Slack bot tokens

   docker run -d --restart=unless-stopped \ --name ops \ -v ops-home:/home/agent \ -e KERN_NAME=ops \ -e OPENROUTER_API_KEY=sk-or-... \ -e SLACK_BOT_TOKEN=xoxb-... \ -e SLACK_APP_TOKEN=xapp-... \ ghcr.io/oguzbilgic/kern-ai
2. 2

   Message the bot; the first person to DM it becomes the operator

3. 3

   Restart later without environment variables, since config is saved in the volume

   docker run -d --restart=unless-stopped -v ops-home:/home/agent ghcr.io/oguzbilgic/kern-ai

Things to know

• !Needs an LLM API key (for example OpenRouter) or a local Ollama setup to run.
• !The README says MIT licensed, but the repository metadata lists no license, so check the license file before relying on it.

Topics

FAQ

What is kern-ai?

A long-running agent runtime that lives in your chat apps (Slack, Matrix, Telegram, Discord, IRC, Nostr) with persistent memory and background jobs. kern runs an AI agent as a long-lived process with a single persistent memory, reachable from Slack, Matrix, Telegram, Discord, IRC, Nostr and a terminal interface. Each agent is configured with a name, a model provider key and a bot account, and runs in its own Docker container with a volume that stores sessions, memory, notes and installed tools.

Is kern-ai free to use? What license does it have?

kern-ai has no declared license. Its code is public, but without a license you have no explicit permission to reuse it; ask the author before relying on it.

Is kern-ai still maintained?

Actively maintained. The last commit to kern-ai was on Oct 8, 2026.

What can I use kern-ai for?

Running an ops or build-monitoring assistant that posts job results into a Slack channel. Giving a team a shared research agent reachable from Telegram or Discord. Operating a self-hosted agent on a homelab with local models through Ollama.

How do I install kern-ai?

Start an agent on Slack with Docker, passing your model key and Slack bot tokens: docker run -d --restart=unless-stopped \ --name ops \ -v ops-home:/home/agent \ -e KERN_NAME=ops \ -e OPENROUTER_API_KEY=sk-or-... \ -e SLACK_BOT_TOKEN=xoxb-... \ -e SLACK_APP_TOKEN=xapp-... \ ghcr.io/oguzbilgic/kern-ai Message the bot; the first person to DM it becomes the operator Restart later without environment variables, since config is saved in the volume: docker run -d --restart=unless-stopped -v ops-home:/home/agent ghcr.io/oguzbilgic/kern-ai

Where can I get kern-ai?

The source code and installation instructions are on GitHub: https://github.com/oguzbilgic/kern-ai