---
name: coo-comeback
description: On-demand sit-rep when Rocky's been away. 5-section comeback-protocol format. Reads last 14 days dailies + all _Log.md + _AgentLog/.
agent: hermes-coo
triggers:
  - "/comeback"
  - "where are we"
  - "what's happened"
  - "sit rep"
  - "catch me up"
delivery: telegram:7377499346
---

# coo-comeback skill

## Behaviour

Run when Rocky asks for a sit-rep after a break (returning from site, weekend off, holiday, sick day).

### Step 1 — read

Read in parallel:
1. Last 14 days of `01_Daily/{date}.md`
2. All `_Brain/_Log.md` files (`10_RateRight`, `11_OpsMan_LFCS`, `12_SiteOps`, `20_Personal`, `30_Goals`)
3. All `_AgentLog/*-{date}.md` files from last 14 days
4. `_System/HQQueue.md`
5. `_System/COO/Outbox.md`
6. `_System/AgentLog/events.ndjson` (last 14 days slice)

### Step 2 — synthesise per Comeback Protocol

Quoting `_System/CLAUDE.md` verbatim:

> 1. **Stalled** — what's been static and shouldn't be
> 2. **Hot** — what's moving and needs attention
> 3. **Promised, not done** — things Rocky said he'd do but didn't
> 4. **Autonomous activity** — what Hermes/OpenClaw did in his absence
> 5. **Pick up first** — single highest-priority item

Output: terse bullets, no preamble, no padding. Per the protocol.

### Step 3 — extra context COO adds (because it's a CEO not a notebook)

After the 5 sections, append:

```
---
Goals scoreboard (since last comeback):
🎯 {goal1}: {cur}/{tgt} (Δ since last comeback)
🎯 {goal2}: ...
🎯 {goal3}: ...

Goal at risk of missing date: {if any — else "none"}
Suggested re-baseline: {if any goal hasn't moved in >14 days — propose what to change}

Suggested first 30 minutes back: {ordered list of 1-3 small wins}
```

### Step 4 — write + deliver

- Telegram pulse (full text)
- Vault: `_System/COO/Comeback-{YYYY-MM-DD-HHMM}.md`
- Append to `_AgentLog/hermes-coo-{today}.md`
- Append to `dispatch-log.ndjson`

## Detection patterns

### "Stalled"

- `_Log.md` entries that haven't changed in >5 days for projects that should be active
- Inbox REQs with `awaiting-` status >3 days
- Airtable Bids with Status unchanged >5 days during bid period
- PM2 services with status `stopped` >5 days
- Goals.md goals with no measurement update >5 days

### "Promised, not done"

Grep last 30 days of dailies + `_Log.md` for patterns:
- "I'll {verb}" without matching "Done: {same verb}"
- "TODO Rocky" without strikethrough
- "by {date}" where date has passed
- "next week" / "tomorrow" / "this evening" without follow-through
- Verbatim quote with date promised + days overdue

### "Hot"

- Daily notes with >5 timestamped sections in last 3 days (high activity)
- `_Log.md` entries in last 3 days (active project)
- Airtable Bids with Status moving in last 3 days
- Kanban tasks created in last 3 days

### "Autonomous activity"

- `_AgentLog/hermes-*-{date}.md` files in last 14 days
- `dispatch-log.ndjson` entries in last 14 days
- `events.ndjson` entries from agents (filter `agent != "rocky"`)

### "Pick up first"

Pick the ONE item that:
- Is highest priority (per goals at risk)
- Smallest (≤30 min)
- Most unblocking (other things wait on it)
- Most easily verified done

## Failure modes

- No daily notes in last 14 days → flag "vault dormant — health check needed"
- All goals at "unknown" → flag "measurement sources unreachable — fix before goals are meaningful"
- No agent activity → flag "agent fleet idle — verify processes running"

## What good looks like

Rocky reads this in 2 minutes, knows exactly what changed while he was away, has one specific next step, and isn't surprised by anything later.