---
name: decision-gating-status-goes-by-cli-not-shortform
description: "status a peer will ACT on (land, hold, box handoff) must go by `spt send` CLI; single-target shortform tags to doyle went unseen twice 2026-09-25 and he made two wrong calls on the silence"
metadata:
  node_type: memory
  type: feedback
  originSessionId: af3c15a8-8d3a-4264-9c61-396b287bde70
  modified: 2026-09-25T03:07:11.247Z
---

Two single-target `@<doyle …@>` shortform notes (02:20Z "IR-155 done, box back, PR #276 pending"; 02:40Z "#276 green, ff-able") never reached doyle. He then (a) ordered a HOLD on a leg that had already finished, reading the idle box as "not launched", and (b) believed #276 was ff-able after main had moved. Both were corrected only when I re-sent by `spt send` (CLI). doyle ruled (CYN5E7IE): "send me status by CLI when it gates a decision I am about to make; I will stop treating tag silence as information."

Distinct from [[multi-target-shortform-drops-every-id-but-the-first]]: these were SINGLE-target, so dropping the second id does not explain them. Cause unmeasured; I did not check DISPATCH_RESULTS for those turns.

**Why:** a peer cannot see a note that did not arrive, and silence reads as a state claim ("not started", "nothing new"). A dropped status costs the RECEIVER a wrong decision, not the sender a retry.

**How to apply:** anything a peer will act on (a land, a hold, a box handoff, "done", a new head sha) goes by `printf … | spt send <id>`, and its `SENT` or `QUEUED` line is the receipt. Shortform is fine only for acks with no decision riding on them. Also say WHO lands in the ack before pushing: #274 crossed with a "held" note the same day ([[a-standing-authorization-that-asks-for-a-checkin-is-not-standing]]).