herdctl: an orchestration layer for Claude Code
herdctl is an orchestration layer for Claude Code. It lets your agents run on a schedule, as part of a fleet, and puts them right in your discord or slack channel.
HerdCTL - Kubernetes for AI Agents
In January 2026 I released HerdCTL, an open-source orchestration layer for Claude Code that enables autonomous AI agent fleets. I built it to power the Egghead Research agent fleet - a collection of specialized agents that work autonomously across my projects without constant human intervention.
The core problem HerdCTL solves is that Claude Code is session-based - you invoke it, it does work, it exits. HerdCTL transforms this into continuous autonomous operation. Agents wake up on cron schedules, respond to Discord messages, and work through GitHub Issues, Jira, or Linear queues automatically. The tagline "Kubernetes for AI agents" captures the vision: declarative YAML configuration, multiple trigger types, and pluggable integrations.
Key features include self-invoking agents that wake on schedules, Discord chat integration for bidirectional communication, session continuity so you can resume any agent job from terminal or chat, and work source integrations that let agents autonomously discover, claim, and complete tasks.
I open-sourced the entire codebase because I think it represents an important shift in how we'll interact with AI coding assistants. The blog posts below cover the technical details - from the architecture decisions to practical guides on building your own agent fleet.
Recent blog posts about HerdCTL
All the posts below are about HerdCTL.