Agentspace – long-running YOLO agent sessions in Docker
Hi HN, I built agentspace because I kept seeing tmux recommended for keeping Claude Code sessions alive over SSH. I find multiplexers painful because they subtly change shell behavior in ways I always forget. They also do not offer a standardized way to run highly permissive agents in a safe way. I also tried worktrees, but it just feels awkward to manage work in that way. Docker can help with all of this, so I thought up a workflow and used some shell aliases to automate spawning Claude/Codex agents into docker containers. I ended up creating a cli to wrap all these methods and would…
What it does
In the maker’s words, at launch
Hi HN, I built agentspace because I kept seeing tmux recommended for keeping Claude Code sessions alive over SSH. I find multiplexers painful because they subtly change shell behavior in ways I always forget. They also do not offer a standardized way to run highly permissive agents in a safe way. I also tried worktrees, but it just feels awkward to manage work in that way. Docker can help with all of this, so I thought up a workflow and used some shell aliases to automate spawning Claude/Codex agents into docker containers. I ended up creating a cli to wrap all these methods and would like to share it. The problems agentspace is solving: - You can run many tasks in parallel all working on their isolated copy of the code. - No death by tapping y. You work on an isolated copy of your code in a fairly safe environment always so it becomes easy to get out of the way and just let the agent do its thing without potentially compromising your machine. Agents spawned by agentspace run containerized in yolo mode by default. - No multiplexers. I have never had a use case for them so I am not used to them at all. The container is just a normal shell. Detach with Ctrl-P Ctrl-Q (standard Docker), reattach with `agentspace my-task attach`. No new muscle memory required. - No git worktrees either. All approaches using this to run parallel agents just never stuck with me. - Sessions typically die on SSH disconnect. Containers always keep running. Reconnect whenever you want. I am now using this daily for running my agent sessions both on my own machine as well as on a remote VM. Features: - Spawn many agents which all run in parallel and work on their own code by mounting a volume to a docker container - Preview results with a dedicated preview command which launches a container attached to the same volume so you can preview changes in real time - Docker support inside the container via a docker-in-docker sidecar (Please note: `--docker` uses `--privileged` under the hood (required for DinD), so it's not a sandbox against actively hostile code. Use it for repos you'd trust on your machine anyway.) - You handle the git commit, push etc. cycle so you stay in control of what goes into your repo. This is enforced at the image level via `PreToolUse` hooks for Claude Code and exec policy rules for Codex, so it holds even under `--dangerously-skip-permissions`. I don't like agents running changes for me, so I want to be in control here Once pushed, it's just a normal branch so you can merge it through your usual PR flow. It does require a git remote though at the moment as this is the only way to get the work out of the container via the agentspace cli (of course you can use any direct docker command as well). Current limitations: - Base image is `node:24-slim`. Node/TypeScript works out of the box, other runtimes need you to extend the image yourself (You can create your own Dockerfile from agentspace and point `AGENTSPACE_IMAGE` at it). I am planning to just include more languages and runtimes in the agentspace base container as well, but so far this works for me. - Supports Claude Code and Codex today. I would like to support Opencode and Pi later - Windows unverified (should work as this project basically only needs node, git and docker, but it hasn't been tested) - Some parts lightly tested or untested. I basically built it last week to scratch my own itch. It has some rough edges for sure. Very curious to get other experiences.
Does the same job
all alternatives →
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, June 2026
the whole month →
Fundraisly▲1,544AI fundraising agent that finds investors and books meetings
AI · Jun 2026 · fundraisly.com
- H6Homebrew 6.0.0▲1,481
Today, I’m proud to announce Homebrew 6.0.0. The most significant changes since 5.1.0 are a new tap trust security mechanism, the new faster, smaller, default internal Homebrew JSON API, sandboxing on Linux, better defaults informed by our user survey, many brew bundle improvements, improved performance and initial support for macOS 27 (Golden Gate). Happy to discuss any questions here!
Dev tools · Jun 2026 · brew.sh
- PU
hope you enjoy
Life & fun · Jun 2026 · vorpus.github.io
- IM
Life & fun · Jun 2026 · hackernewstrends.com