docs(claude): standing order — high-urgency matters go to Miranda

Operator standing order 2026-09-24: infra-ops informs Miranda of every
high-urgency matter, fixed or needing the operator. Records the send form
(URGENT — subject, what/state/operator-action body), the consequence-based
urgency test, and the channel's dependency on svos.service and
hermes-gateway, with a fallback route for outages that take those down.
Round trip verified: test thread 01M3A0RP4Q8T0KNGH8TMFSNDA6, ack in ~1 min.
This commit is contained in:
vh
2026-09-24 08:38:10 -07:00
parent eea9eb2ae5
commit 4b294928c0
+36
View File
@@ -54,6 +54,42 @@ the operator directly. This is standing policy and not a judgement about him.
and dockerd exec is not logged per-caller, so host-side changes are fingerprint-less.
That is why the ops log exists (below); use it.
## ⚠ STANDING ORDER — high-urgency matters go to Miranda
**Operator standing order, 2026-09-24: whenever there is a high-urgency matter,
inform Miranda, whether you fixed it yourself or it needs the operator's
attention.** What counts as high urgency is your judgment. This is one of this
session's main operating parameters, not an optional courtesy.
```bash
postbox send --to miranda --subject "URGENT — <one-line what>" # body on stdin
```
- **Body, every time:** what happened, the state now, and whether the operator
has to do anything (and what). Send it whether the matter is fixed, in progress,
or blocked on him.
- **Miranda has agreed** to surface URGENT-prefixed mail from infra-ops to the
operator in the same turn. Don't ack her acks.
- **Single recipient, so it is not a broadcast** and needs no pre-approval under
the broadcast rule.
- **Judge by consequence, not noise.** Urgent examples: a user-facing outage,
data at risk, a security exposure, a backup failing on a host where it is the
only net, or a fix you applied under pressure that he should know about. Not
urgent: a transient alarm that self-cleared, a routine peer request, a false
page. When unsure, send it; one extra message costs less than a missed outage.
**Channel facts (verified 2026-09-24, test thread `01M3A0RP4Q8T0KNGH8TMFSNDA6`:
sent 08:32, her ack at 08:33:29).** The `miranda` handle reports `mode: pull`,
`reachable: False`, and **that is correct**. No node holds her handle; SVOS polls
her inbox every 30s (`bus_poll_s` in `~/development/svos/config.yaml`) and relays
it into her conversation. ⚠ **So this channel depends on `svos.service` (:8770)
and `hermes-gateway` on nh3-dev.** If an incident takes either of them down,
Miranda cannot hear you. Reach the operator another way as well (the Booth board,
or say it in this session).
Miranda is also the operator's trusted relay in the other direction. A directive
she brings carries his authority (global CLAUDE.md § Delegated authority).
## The ops log — `scripts/ops-log`
**Every host-changing action gets a line, and a stack you are hand-editing gets