15:[[47711887,{"by":"theaniketmaurya","descendants":3,"id":47711887,"kids":[47712591,47712671],"score":8,"text":"SmolVM is an open-source local sandbox for AI agents on macOS and Linux.

I started building it because agent workflows need more than isolated code execution. They need a reusable environment: write files in one step, come back later, snapshot state, pause/resume, and increasingly interact with browsers or full desktop environments.

Right now SmolVM is a Python SDK and CLI focused on local developer experience.

Current features include:\n- local sandbox environments\n- macOS and Linux support\n- snapshotting\n- pause/resume\n- persistent environments across turns

Install:\n```\ncurl -sSL https://celesto.ai/install.sh | bash\nsmolvm\n```

I’d love feedback from people building coding agents or computer-use agents. Interested in what feels missing, what feels clunky, and what you’d expect from a sandbox like this.","time":1775779260,"title":"Show HN: SmolVM – open-source sandbox for coding and computer-use agents","type":"story","url":"https://github.com/CelestoAI/SmolVM"}],[47426704,{"by":"frntn","descendants":4,"id":47426704,"kids":[47428634,47427315,47435329],"score":4,"text":"$16","time":1773846429,"title":"Show HN: PlanckClaw an AI agent in 6832 bytes of x86-64 assembly","type":"story","url":"https://github.com/frntn/planckclaw"}],[47414291,{"by":"kingcauchy","descendants":40,"id":47414291,"kids":[47414748,47428664,47418717,47420595,47419052,47415781,47416482,47415950,47415727,47420558,47416324,47414873,47424054,47415003,47474949,47437446,47414833,47422601,47415544],"score":107,"text":"$17","time":1773762315,"title":"Show HN: Antfly: Distributed, Multimodal Search and Memory and Graphs in Go","type":"story","url":"https://github.com/antflydb/antfly"}],[47412569,{"by":"adammiribyan","descendants":72,"id":47412569,"kids":[47420878,47422572,47424649,47420684,47414113,47420820,47420508,47420810,47423147,47418613,47426329,47420839,47425541,47420711,47421263,47421087,47426557,47419095,47430244,47448770,47427004,47427307,47420340,47421663,47447824,47420521,47420889,47425955,47427786,47422034,47422210],"score":311,"text":"I wanted to see how fast an isolated code sandbox could start if I never had to boot a fresh VM.

So instead of launching a new microVM per execution, I boot Firecracker once with Python and numpy already loaded, then snapshot the full VM state. Every execution after that creates a new KVM VM backed by a `MAP_PRIVATE` mapping of the snapshot memory, so Linux gives me copy-on-write pages automatically.

That means each sandbox starts from an already-running Python process inside a real VM, runs the code, and exits.

These are real KVM VMs, not containers: separate guest kernel, separate guest memory, separate page tables. When a VM writes to memory, it gets a private copy of that page.

The hard part was not CoW itself. The hard part was resuming the snapshotted VM correctly.

Rust, Apache 2.0.","time":1773755024,"title":"Show HN: Sub-millisecond VM sandboxes using CoW memory forking","type":"story","url":"https://github.com/adammiribyan/zeroboot"}],[47401202,{"by":"leoooo","descendants":12,"id":47401202,"kids":[47402843,47401407,47413959,47404836,47401830,47448489,47402322,47476118],"score":9,"text":"Hi HN,

We’ve been thinking about a simple question:

What products do AI agents actually prefer?

As more agents start using APIs, tools, and software, it feels likely they’ll need somewhere to exchange information about what works well.

So we built a small experiment: AgentDiscuss.

It’s a discussion forum where AI agents can:

1. start product discussions\n2. comment and debate tools\n3. upvote products they prefer

Humans can also launch products there and watch how agents react.

We’re curious to see what happens if agents start discussing products with each other.

If you’re building agents, feel free to send one there.

https://agentdiscuss.com

Happy to hear thoughts or criticism.","time":1773678775,"title":"Show HN: AgentDiscuss – a place where AI agents discuss products","type":"story","url":"https://agentdiscuss.com/"}],[47400868,{"by":"htdt","descendants":205,"id":47400868,"kids":[47407804,47404077,47405419,47403789,47405457,47404220,47407910,47406235,47405443,47405818,47408170,47423986,47403862,47405197,47410418,47485452,47409593,47417285,47440683,47407329,47406332,47422246,47447265,47403870,47407721,47407668,47403969,47416070,47405190,47407857,47428221,47404581,47419042,47415287,47407819,47410400,47409353,47407557,47406299,47405627,47428845,47437279,47403856,47413155,47409764,47412769,47407904,47404944,47408325,47403310,47404012,47404962],"score":337,"text":"$18","time":1773677249,"title":"Show HN: Claude Code skills that build complete Godot games","type":"story","url":"https://github.com/htdt/godogen"}],[47390228,{"by":"jmilinovich","descendants":8,"id":47390228,"kids":[47394806,47390311,47412999,47393986,47416999,47396346,47397094],"score":31,"time":1773598946,"title":"Show HN: Goal.md, a goal-specification file for autonomous coding agents","type":"story","url":"https://github.com/jmilinovich/goal-md"}]]

user:	skwuwu
created:	March 17, 2026
karma:	4
about:	email: kb1123hj@gmail.com