AI / Agentic

Show HN: Ductwork – A Go platform for running AI agents on autopilot

HackerNews · dneil86753/1/2026, 10:41:27 PM 4 points 3 comments
View Original Source

Summary

I&#x27;ve been running Claude agents for various automation tasks — monitoring crypto news, syncing Todoist, running health checks — and I kept hitting the same problem: there&#x27;s no clean way to deploy an agent that just runs on a schedule without a human babysitting it.<p>Every agent framework I looked at was built around chat interfaces or one-shot workflows. I wanted something closer to cron for AI agents — define a task, give it a schedule, let it run forever. So I built Ductwork.<p>You define tasks as simple JSON files — a prompt, a schedule, optional memory and skills — and ductwork handles scheduling, execution, retries, and history. The agents have bash, file read&#x2F;write, and that&#x27;s it. No fancy abstractions.<p>The thing that makes it actually useful for unattended operation:<p>Persistent memory — agents write to a memory directory between runs. My Bitcoin news monitor remembers which articles it&#x27;s already reported on. Next run, it only flags new ones.<p>Security boundaries — if you&#x27;re letting agents run unsupervised, you need guardrails. Per-task tool whitelists, path restrictions, bash command filters. A monitoring task can&#x27;t accidentally rm -rf something.<p>Run history and observability — every run is tracked with status, duration, token usage, and errors. REST API for everything so you can integrate with whatever alerting you already use.<p>It scales from a single process (ductwork start) to distributed — same binary with --mode=control runs a task queue, --mode=worker on other machines polls for work. No new dependencies, just HTTP.<p>Single Go binary, go install and you&#x27;re running. ~3,500 lines, only deps are the Anthropic SDK and Cobra.<p>This is definitely not a finished product — it&#x27;s early and there&#x27;s a lot I want to add. But it&#x27;s functional and I&#x27;d love for people to download it, play around with it, and let me know what they think. Feedback, ideas, issues — all welcome.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;dneil5648&#x2F;ductwork" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;dneil5648&#x2F;ductwork</a>

Tags

#ai#claude#agent#github

Metadata

Article ID
#212
Source
HackerNews
Scraped At
3/2/2026, 7:10:05 AM
URL Hash
2a1f57bc9da43ee9…