read.internal.disk_usage

How full this box’s disk is, as the kernel’s disk alarm last measured it (every minute): state is ok, high (at or above threshold_percent, FORGE_DISK_ALARM_PERCENT, default 85) or unknown — never ok when the measurement failed, has not run yet or is stale; reason then says which. used_percent follows df. A fact about the BOX, the same for every tenant on it; the admin draws a warning band from it and the kernel log carries one disk_alarm state=… line per transition. Internal face only.

• Surface: GET /v1/read/internal/disk_usage · MCP tool read.internal.disk_usage
• Auth: ★ Internal read (operator) — a tenant credential is REQUIRED; requires the tenant.settings.read scope. The tenant is resolved from the CALLER’s identity, never from store. Personal fields come back MASKED for an actor without pii.read (the shape is unchanged).

Params (JSON Schema)

{
  "properties": {},
  "type": "object"
}

Result (200)

Open (unknown) — this capability does not declare an output schema.

Errors

invalid_params (400) · not_found (404) · unauthorized (401) · forbidden (403)