Documentation
Everything you need to create, configure, and run kern agents.
Get Started
Install kern, create your first agent, and start chatting in under a minute.
Docker
Deploy coworker agents with Docker and Compose — volumes, environment, and lifecycles.
Architecture
Directory-based agent model, multi-channel session queue, and process isolation.
Configuration
Agent config, environment variables, model selection, and provider options.
CLI Reference
Commands — init, run, status, tui, web, backup, restore, and scripts.
Chat Commands
In-chat slash and bang commands — !status, !wyd, !jobs, !subagents, !skills, !restart.
Interfaces
Slack, Matrix, Telegram, Discord, IRC, Nostr — how agents connect to every channel.
Tools
Built-in tools — bash, jobs, spawn, subagents, platform APIs, read, write, edit, grep.
Sub-Agents
Parallel delegated research workers with initiator-aware return routing.
Memory
How agents remember — topic DAG, vector recall, and git-backed knowledge.
Context
Prompt assembly, token budgeting, segmentation, and summary rollups.
Caching
Prompt caching — breakpoints, stable trim boundaries, and cost reduction.
Skills
AgentSkills universal format — dynamic in-chat activation without restarts.
MCP
Model Context Protocol — extend toolboxes with external MCP servers.
Media
Multi-modal vision pre-digest, PDF extraction, and voice message audio I/O.
Pairing
Operator pairing codes, access control, and user permission tracking.
Quick start: npm install -g kern-ai && kern init my-agent