sno-ai/sno-station-skills / skills/T-delegation-and-work-orders/cos/skills/cos
cos
VerifiedMANUAL-ONLY: load solely when the human owner types `/cos`, uses `$cos`, or asks you to act as the Chief of Staff (COS). The COS is the one agent the owner talks to; it coordinates the owner and the per-project Project Leads (PLs). A task card, charter, or dispatch makes you an executor, not COS.
| File | Kind | Size |
|---|---|---|
| LEARNING.md | markdown | 6.1 KB |
| SKILL.md | skill md | 76.0 KB |
| references/card-template.md | markdown | 10.9 KB |
| references/digest-template.md | markdown | 4.9 KB |
| references/owner-profile.md | markdown | 5.3 KB |
| references/pocket-card.md | markdown | 1.6 KB |
| scripts/cos-claim.shexecutable | script | 31.4 KB |
| scripts/cos-claim.texecutable | script | 13.4 KB |
| scripts/cos-roster.shexecutable | script | 40.5 KB |
| scripts/cos-roster.texecutable | script | 7.6 KB |
| scripts/cos-screen.shexecutable | script | 4.8 KB |
| scripts/cos-screen.texecutable | script | 2.5 KB |
| scripts/cos-since.shexecutable | script | 3.6 KB |
| scripts/exec-sentinel.shexecutable | script | 6.1 KB |
| scripts/exec-sentinel.t | other | 2.4 KB |
| scripts/exec-tree-check.shexecutable | script | 2.6 KB |
| scripts/selftest/run-selftest.shexecutable | script | 204 B |
| scripts/selftest/text-check.py | script | 1.0 KB |
scripts/cos-screen.sh
View on GitHub#!/usr/bin/env bash
# cos-screen.sh — print what a PL's window shows RIGHT NOW, so a status answer to the
# owner comes from the live screen and not from inbox and disk inference.
#
# cos-screen.sh <seat-address | repo> [--lines N] (default 40 lines)
#
# seat-address e.g. pl.myrepo@host1 — the window that seat registered.
# repo a registry home_repo — every live-listed PL seat of that repo;
# if none of them can be read, every Orca terminal or tmux pane
# whose directory basename equals the repo name.
#
# tmux windows are read with `tmux capture-pane -p` on the default tmux server. Orca
# terminals, if Orca is installed, are read with `orca terminal read --screen` (the
# rendered frame; the default stream mode returns repainted lines as stacked fragments).
# Needs tmux, jq and the sno command.
#
# EXIT: 0 at least one screen printed · 2 usage · 3 no window could be read (a line
# starting NO WINDOW says why — that is a finding, not a failure to retry).
set -Eeuo pipefail
REGISTRY="${SNO_PL_REGISTRY:-$HOME/.local/state/pl-registry.tsv}"
lines=40
target=''
printed=0
die() { printf 'cos-screen: %s\n' "$*" >&2; exit 2; }
for tool in tmux jq sno; do
command -v "$tool" >/dev/null 2>&1 || die "$tool is required but not installed"
done
while (($#)); do
case "$1" in
--lines) [[ "${2:-}" =~ ^[0-9]+$ ]] || die "--lines needs a number"; lines="$2"; shift 2 ;;
-h|--help) sed -n '2,18p' "$0"; exit 0 ;;
-*) die "unknown flag: $1" ;;
*) [[ -z "$target" ]] || die "one target only"; target="$1"; shift ;;
esac
done
[[ -n "$target" ]] || die "usage: cos-screen.sh <seat-address | repo> [--lines N]"
# capture-pane and Orca pad the frame with blank rows; keep the last N non-blank rows.
last_lines() {
awk -v n="$lines" 'NF { last = NR } { l[NR] = $0 }
END { for (i = (last > n ? last - n + 1 : 1); i <= last; i++) print l[i] }'
}
show() { # header, channel, handle-or-pane
local header="$1" channel="$2" handle="$3" json
printf '=== %s [%s]\n' "$header" "$channel"
case "$channel" in
orca)
json="$(orca terminal read --terminal "$handle" --screen --json 2>/dev/null)" || true
if jq -e '.ok' >/dev/null 2>&1 <<<"$json"; then
jq -r '.result.terminal.tail[]' <<<"$json" | last_lines
printed=1
else
printf 'NO WINDOW: orca terminal %s not readable (%s)\n' "$handle" \
"$(jq -r '.error.code // "no reply"' <<<"$json" 2>/dev/null || echo "no reply")"
fi ;;
tmux)
if json="$(tmux capture-pane -p -t "$handle" 2>/dev/null)"; then
last_lines <<<"$json"
printed=1
else
printf 'NO WINDOW: tmux target %s not found on the default server\n' "$handle"
fi ;;
*) printf 'NO WINDOW: channel %s has no screen to read\n' "$channel" ;;
esac
}
by_seat() {
local address="$1" seat channel handle state
seat="$(sno reach seats --json 2>/dev/null | jq -c --arg a "$address" 'select(.address == $a)' | head -n1)"
if [[ -z "$seat" ]]; then
printf '=== %s\nNO WINDOW: no registered seat with this address\n' "$address"
return
fi
channel="$(jq -r '.channel' <<<"$seat")"
handle="$(jq -r '.handle' <<<"$seat")"
state="$(jq -r '.state' <<<"$seat")"
# tmux seat handles read tmux-<server-hash>:<session>; the session is the target.
[[ "$channel" != tmux ]] || handle="=${handle##*:}:"
show "$address (seat $state)" "$channel" "$handle"
}
by_directory() {
local repo="$1" handle title pane path
printf 'no seat window readable for %s; looking for windows whose directory is named %s\n' "$repo" "$repo"
if command -v orca >/dev/null 2>&1; then
while IFS=$'\t' read -r handle title; do
show "$title" orca "$handle"
done < <(orca terminal list --json 2>/dev/null |
jq -r --arg r "$repo" '.result.terminals[]? | select((.worktreePath | split("/") | last) == $r) | [.handle, .title] | @tsv')
fi
if command -v tmux >/dev/null 2>&1; then
while IFS=$'\t' read -r pane path title; do
if [[ "${path##*/}" == "$repo" ]]; then show "$title" tmux "$pane"; fi
done < <(tmux list-panes -a -F '#{pane_id} #{pane_current_path} #{session_name}: #{pane_title}' 2>/dev/null || true)
fi
}
if [[ "$target" == *@* ]]; then
by_seat "$target"
else
while IFS= read -r address; do
by_seat "$address"
done < <(awk -F'\t' -v r="$target" '$1 == r && $3 ~ /^pl\./ && $7 != "RETIRED" { print $3 }' "$REGISTRY" 2>/dev/null || true)
[[ "$printed" == 1 ]] || by_directory "$target"
fi
if [[ "$printed" == 0 ]]; then
printf 'NO WINDOW: nothing readable for %s — treat it as having no window\n' "$target"
exit 3
fi
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |