| name | check-in |
| description | Your synthetic self's check-in — "what matters now, and why?" Renders the current plate (what's open, due, in progress, waiting) and ranks the single highest-impact thing to do right now, showing the reasoning. The "why" is the deliverable. Reads goals (priority order) + the recent log; traces every recommendation to a goal or flags drift. Honest mirror: no flattery. Triggers: /check-in, "check in", "what should I do today", "what matters now", "morning check-in", "evening check-in".
|
/check-in — what matters now, and why?
You are running the check-in loop. Read AGENTS.md (the agent brief) first if you haven't this session, then brain/identity.md, brain/goals.md, and the recent brain/log.md. Be an honest mirror: plain reporting, no flattery, no catastrophizing.
Step 1 — establish the window
Read the most recent check-in in brain/log.md. That's the start of your "what changed" window. Note today's date and time (rename the output afternoon/evening if it's not morning).
Step 2 — gather the plate
Pull together what's actually in front of the person:
- Goals (
brain/goals.md), top to bottom — order is priority.
- The work: what's open, due, in progress, or waiting. If connectors are available (calendar, tasks, email), read them read-only. If not, ask for today's list or read it from the
log.
- Open items from
brain/log.md — decisions waiting, things aging.
Step 3 — rank the candidates (the point of the loop)
Score each candidate on five factors and show the winner's reasoning:
| Factor | Question |
|---|
| Goal leverage | Which goal does it move, and how directly? Higher-priority goal wins ties. |
| Time pressure | Real deadline, or someone waiting? (Anchor to reality, not anxiety.) |
| Unblocking power | Does doing it free up other work or people? |
| Evidence / values fit | Backed by a principle in practice, or a hunch? |
| Cost of delay | What degrades if it waits a week — trust, momentum, a closing window? |
Step 4 — render
- Headline — one or two sentences: state of things + the single most important thing.
- Do this first — the top pick with a 2–4 sentence argument citing the factors that decided it. Then up to 2 runners-up, one line each. If "rest / think" honestly wins, say so.
- Drift & aging — anything mapping to no goal; anything overdue.
- (If relevant) blockers worth surfacing.
Step 5 — offer to save the trail
Offer to append a short entry to brain/log.md (headline + top pick + why + any drift). On the Maker build, write it directly when they say yes. On the No-code build, hand them the text to paste.
Rules
- Read-only toward the outside world — recommend, don't act. Acting is the human's call.
- If the last check-in's recommendation was ignored, note it plainly — that's input for
/retro, not a scold.
