Observability
See what every sandbox uses and what it costs. CPU, memory, disk and usage, all in one place.
CPU & memory ⌄
1h6h24h7d30d
All sandboxesCPUMemory
Cost
By hour$4.16 today
Disk
Disk used36 GiB of 40, at most
Active sandboxes
12
running now
28 total today↑ 12% vs yesterday
See what it uses.
Each sandbox's CPU, memory, disk and network are read once a minute while it runs, and kept for 15 days.
- Read from outside the virtual machine. Nothing is installed in the sandbox, and nothing running in it is asked.
- One call for a sandbox over any stretch of time, and one for all of your sandboxes at one moment.
- Each reading comes with the size it is a share of, so a sandbox that would fit a smaller machine is plain to see. The small one is $0.027 an hour.
- The console draws every sandbox with what it reads, and moves them all back through the last 24 hours.
- A stopped sandbox reads nothing.
CPUMemoryNetwork
Run a fleet, and see what each agent costs.
Every stretch a sandbox runs is recorded, and read back by sandbox, by key, by day and by hour.
- Counted by the second, and priced at the list price.
- For any stretch of time. Without one, the month so far, and where it ends at that pace.
- Narrowed to one key or several, mail and all.
- A sandbox deleted since still counts for the time it ran.
KeySandboxesHoursCost
ci-runner14212.5$5.74
review-agent396.0$2.59
research141.2$1.11
Watch them where you watch everything else.
Every minute, each running sandbox's CPU, memory, disk and network go to your own OpenTelemetry collector, and each start, stop and error as it happens.
- One OTLP/HTTP address and its key: Datadog, Grafana Cloud, Honeycomb, or a collector you run.
- A test export as soon as it is set, so that a wrong key shows at once and not a minute later.
- Network and disk bytes a second, as well as CPU, memory and disk.
- Each event has the same id in your collector as in your webhooks, so a duplicate is plain to see.
- When your collector was down, what it missed is sent once it is back: up to 10 minutes of readings and 24 hours of events.
EndpointOTLP over HTTPS, your address
MetricsCPU, memory, disk, network, each minute
Eventsstarted, stopped, moved, failed
Do the same by hand, in the console.
A console in your browser, for the times a person needs to look at what the agents are doing.
- Create, start, stop and delete sandboxes, and run a command in one.
- Read and write files, open ports, give it addresses of its own and take snapshots from a sandbox's page.
- Make templates, keep their secrets and set what their sandboxes may reach, and make and delete API keys.
- Usage by day, by sandbox and by key, and the plan you are on.
- Every sandbox drawn with its CPU, memory and disk, now or at any moment of the last 24 hours.
- Set the collector your metrics and events go to, and see how its last export went.
- Add webhooks, pick what each is sent, and see how every post went.
sandbox-01RunningsmallCPU 42%$0.027 / hour
data-extractionRunningmediumCPU 67%$0.054 / hour
marketing-agentStoppedsmall—$0.0014 / hour
Also part of Observability:Webhooks, written up under Connectivity
The five pillars
All features, on one page→Give your agent somewhere to work.
Free to start, with +$10 credit.