logs

logs

crabbox logs prints the retained command output for a recorded run.

crabbox logs run_abcdef123456
crabbox logs --id run_abcdef123456
crabbox logs run_abcdef123456 --tail 80
crabbox logs run_abcdef123456 --json

Run IDs have the form run_<hex>. Get them from history, or from the run.started event in events.

What gets stored

When crabbox run executes against a coordinator, it streams the remote process stdout and stderr to your terminal and records a bounded copy on the coordinator. The retained log is the combined command output, capped at 8 MiB per run; the coordinator stores it in 64 KiB chunks so a noisy parallel run does not blow past Durable Object storage limits.

Output beyond the cap is dropped and the run is flagged as truncated, so a consumer can tell the tail is missing.

Runs started with run --capture-stdout <path> intentionally omit stdout from the retained log and from output events; stderr is still streamed and retained. Use that mode for binary stdout, Sixel frames, archives, or anything that should not be stored as text on the coordinator. Files pulled with run --download remote=local are local artifacts and never appear in coordinator logs.

Output

The default form writes the log text to stdout, unmodified. Add --tail N to print only the last N lines of the retained log.

Logs are stored as combined output; stdout and stderr are not separately indexed in the retained log, so per-stream filtering belongs on events instead.

--json prints the full run record alongside the log:

{
  "run": {
    "id": "run_abcdef123456",
    "leaseID": "cbx_abcdef123456",
    "state": "succeeded",
    "exitCode": 0,
    "logBytes": 4096,
    "logTruncated": false
  },
  "log": "..."
}

The run object carries the same fields you get from history (provider, target, timings, exitCode, logBytes, logTruncated, parsed results, and more); the log field holds the retained text after any --tail trim. Scripts that want exit code plus log text in one payload should read run.exitCode and log.