Hacker News: iamsky

New comment by iamsky in "Show HN: Vibe zsh, turn natural language into shell commands"

iamsky — Tue, 30 Jun 2026 06:49:06 +0000

It definitely can, it just depends on the complexity of the commands you want it to generate. Most of the time a small local model will do what you want. If you know of a small local model fine tuned to generate shell commands I'm all ears. I usually run it with a 9B qwen model.

Pairing this with dictation is fun though, and speeds things up a lot for terminal heavy work.

Show HN: Vibe zsh, turn natural language into shell commands

iamsky — Mon, 29 Jun 2026 23:01:22 +0000

just type what you want to do (or use a dictation tool) press ctrl+g and voila, you have a command.

Comments URL: https://news.ycombinator.com/item?id=48726461

Points: 3

# Comments: 3

Show HN: Codemcp, One MCP tool that lets agents write Python to call many

iamsky — Wed, 24 Jun 2026 11:41:30 +0000

See the readme. 55-90% input token savings. You can run this test yourself in the /bench directory.

Comments URL: https://news.ycombinator.com/item?id=48658274

Points: 4

# Comments: 0

New comment by iamsky in "DjinnBot – Deploy a full autonomous AI engineering team in one curl command"

iamsky — Sun, 22 Feb 2026 22:29:44 +0000

I got tired of spending days wrestling with fragile multi-agent setups only to get generic, uncoordinated agents that didn't feel like a real team. So I built DjinnBot — a containerized, opinionated platform that spins up a complete autonomous AI software team in under 5 minutes:

Product Owner (Eric — pushes back on vague specs) Solutions Architect (Finn) Senior SWE, UX, Test Engineer, SRE + on-demand specialists They spec, design, code, review PRs, test, and deploy — fully collaboratively

What makes it different:

One curl command + wizard (Docker, secrets, optional SSL — everything handled) Each agent runs in its own isolated container with full toolchains (Node, Python, Go, Rust, git, etc.) Persistent semantic memory (ClawVault) so the team actually improves over time Strong personas — agents debate, critique, and reject bad ideas Beautiful real-time dashboard with Kanban board, live streaming output, and pipeline view Every agent is a native Slack bot (threaded discussions) YAML pipelines for custom workflows + MCP proxy for external tools (GitHub, web search, etc.)

Completely free and open-core to self-host (FSL-1.1-ALv2 license → Apache 2.0 after 2 years). No Kubernetes, no cloud accounts required. I'm dogfooding it hard — the AI team is currently helping me build and improve DjinnBot itself. Would love honest feedback, especially if you've tried other agent frameworks (OpenClaw, etc.). What works? What feels off? Any projects you'd throw at it? Docs & 1-click install: https://docs.djinn.bot GitHub: https://github.com/BaseDatum/DjinnBot Cloud waitlist: https://app.djinn.bot Thanks!

DjinnBot – Deploy a full autonomous AI engineering team in one curl command

iamsky — Sun, 22 Feb 2026 22:29:44 +0000

Article URL: https://docs.djinn.bot/

Comments URL: https://news.ycombinator.com/item?id=47115451

Points: 2

# Comments: 1