sandstorm
Run thousands of AI agent sandboxes on the cheapest machines.
Runs with mockssource: GitHubRustApache-2.0commit d27ab7c45212
Rust, Apache-2.0 licensed. The project labels itself: agent infrastructure, ai agents, checkpoint restore, code execution, gvisor, rust, sandbox and self hosted.
sandstorm runs, with stand-ins for the services it depends on. An Argusic agent installed it in 15 minutes and hit 3 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
Sandstorm release build compiles, web console builds, all 412 tests pass, and the master API serves all endpoints (healthz, CA cert, sandboxes, nodes, templates, Prometheus metrics, and web console) on 127.0.0.1:8080 with --insecure-transport and --api-open. Node mode requires root for iptables/ip netns operations which are unavailable in this container.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 92 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page. Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.
What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- iptables and ip not found: node backend failed with 'No such file or directory'3 minutes
- zstd not installed: 3 tests failed (crates/sandstorm-core/src/node/detach_tests.rs)2 minutes
- /run/sandstorm/ not writable: node mode failed to create /run/sandstorm/gateway-ports and /run/sandstorm/veth1 minute
- Install time
- 15 minutes
- Cold machine to finish
- 12 minutes
- Errors hit and fixed
- 3 hit, 3 fixed with no human help
- How the result was proved
- cargo test -p sandstorm-core exited 0 (412 passed); curl http://127.0.0.1:8080/healthz returned 'ok' (200); curl http://127.0.0.1:8080/ca.pem returned 200 (611 bytes); curl http://127.0.0.1:8080/v1/sandboxes returned [] (200); curl http://127.0.0.1:8080/v1/nodes returned [] (200); curl http://127.0.0.1:8080/v1/templates returned [] (200); curl http://127.0.0.1:8080/metrics returned Prometheus metrics (200); curl http://127.0.0.1:8080/ returned console HTML (200)
- Model tokens used
- 122,040
- Exact commit tested
- d27ab7c45212
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
- Behaviour against real services
- this run was verified against stand-in services only
Watch the session
The whole run, from a clean machine, at the speed it actually happened. Nothing is cut and nothing is sped up.
A recording of attempt 1, run , the attempt that decided the verdict above. The terminal recording it was made from is the evidence, stored with a sha256 fingerprint; the video is a rendering of it. Nothing loads from the video host until you press play.
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
<meta name="viewport" content="width=device-width, initial-scale=1"> The master serves Prometheus metrics and the web console, all endpoints respond correctly. Now let … exec /bin/sh -lc 'echo "Current time:" && date -u +%s && echo "Total minutes estimate: TBD"' in /work/re… succeeded in 0ms: Current time: 1793604431 Total minutes estimate: TBD All verified. Here is the final result. tokens used 122,040 All verified. Here is the final result.
Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Installed in 15 minutes, faster than the median of the 6 comparable projects Argusic has measured.
- Recovered from all 3 errors without a human stepping in, which says the failures are documented well enough to solve.
- Licensed Apache-2.0, as reported by its host.
What did not, or is not known
- Only reached a running state against stand-in services, so nothing here proves it works against the real ones.
- Measured once so far. A single attempt shows it can run, not that it runs reliably.
Of the 7 agent-infrastructure projects Argusic has installed and timed, sandstorm was the 4th fastest to reach a running state, and 6 of 7 reached one at all.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
| tested by | attempt | outcome | score | cost USD | when |
|---|---|---|---|---|---|
| Argusic Runner | 1/3 | Runs with mocks | 92.00 | 0.31 |
Topics (from GitHub)
agent-infrastructureai-agentscheckpoint-restorecode-executiongvisorrustsandboxself-hostedsnapshot
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/sandstorm)Questions
- Does sandstorm run?
- sandstorm runs, with mocks standing in for external services. Argusic installed and launched it on a clean machine in 15 minutes, hitting 3 errors on the way, and recorded the session.
- How did Argusic test sandstorm?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit d27ab7c45212. One attempt is recorded, and the full method is on the methodology page.
- What did it cost to test sandstorm?
- The run that produced this verdict cost $0.31: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does sandstorm take to install?
- 15 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does sandstorm need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing sandstorm?
- 3 things broke in the recorded run, and 3 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does sandstorm compare with the alternatives?
- Of the 7 agent-infrastructure projects Argusic has installed and timed, sandstorm was the 4th fastest to reach a running state, and 6 of 7 reached one at all.
- Where is the evidence for sandstorm?
- The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.