remote-compute-ssh

1This skill covers remote compute over SSH, including direct execution and Slurm submission:
2listing hosts, creating handles, running short remote commands (callCommand), reading/writing host
3knowledge docs, and the full async job lifecycle — submit → save `job_id` → read non-blocking
4snapshots by that ID →
5harvest → analysis turn → publish artifacts.
6 
7**Where host.compute runs:** `host.compute` lives ONLY on the control-plane REPL kernel — run
8every example below with the `repl_execute` tool (JavaScript), the same kernel that hosts
9`host.mcp`. The `python`/`r` data kernels have NO `host.compute` (SSH and approvals stay outside
10the sandbox workspace); calling it from a python/r cell will fail with `host.compute is undefined`.
11 
12## External compute targets
13 
14Use the execution mode returned by `host.compute.listHosts()` to select the file layout and launch options. Each external job requires its own approval, including the bound target, resource request and frozen inputs. SSH-only remote paths and named SSH environments do not apply.
15 
16- `kubernetes`: The user configures the context, namespace, workload image and transfer image. The command runs in `/workspace`, where frozen inputs appear under their `dstFilename`. The app collects outputs from the job volume.
17…