daily-plan

name	daily-plan
description	Morning planning loop. Reads context/Priorities.md, kb/wiki/_hot.md, and last 7 days of decisions/log.md. Proposes ONE top focus for the day and identifies one AI-assisted task. Read-only — does not modify anything except (optionally) decisions/log.md with a single-line entry.
disable-model-invocation	true

/daily-plan — Morning Loop

First skill of the day. Aligns the day's focus with the week's priorities before the inbox steers things.

When to use

• First thing in the morning, before opening email/Slack
• After a long break (returning from PTO, weekend, etc.)

When NOT to use

• Multiple times per day — defeats the purpose
• When context/Priorities.md is still placeholder — run /onboard first

Inputs

• context/Priorities.md
• kb/wiki/_hot.md (if exists and < 7 days old)
• decisions/log.md last 7 days
• (Optional) calendar / inbox snapshot — if MCP is connected

Outputs

• Three-line summary in chat:
  • Top focus: one sentence
  • AI-assisted task: which slash command + what
  • What I'm not doing today: explicit non-goals
• (Optional) one-line entry in decisions/log.md: ## YYYY-MM-DD — Day plan: <focus>

Hard rules

• MUST be terse. Three lines, not three paragraphs.
• MUST NOT propose more than one top focus. The constraint is the value.
• MUST flag when _hot.md is stale (>7 days) — but don't refuse to run.
• MUST NOT modify context/, connections.md, or any skill/agent.

Steps

1. Read context/Priorities.md. Identify "this week" focus.
2. Read kb/wiki/_hot.md. Note: leverage gap of the week, hot articles, candidates.
3. Read decisions/log.md last 7 days. Note: recent commitments, pending follow-ups.
4. Synthesize:
   • Top focus: what advances the week's priority most?
   • AI-assisted task: what part can Claude do or accelerate?
   • Not doing today: name 1-2 things you're explicitly punting
5. (Optional) Append a one-line entry to decisions/log.md.

Failure modes

Symptom	Cause	Fix
Priorities.md is placeholder	Onboarding incomplete	Pause; suggest /onboard
_hot.md is stale	Last weekly review was >7 days ago	Note the staleness; run /weekly-operating-review later today
Three+ candidate focuses	Real ambiguity in priorities	Propose ONE; surface the others as "not doing today"
No AI-assisted task is obvious	Day is admin-heavy	That's fine. Say so — "no AI lift today" is a valid output.

Reference files

• context/Priorities.md
• kb/wiki/_hot.md
• references/3ms-framework.md — Find Constraint
• docs/CADENCE.md

Improvement loop

After each run:

1. Did the operator follow the proposed top focus? If consistently no → recommendation logic is wrong
2. Did the AI-assisted task actually get used? If not → wrong slash command suggested
3. Did the day end where it started (per /end-of-day-review)? If yes for 3+ days → priorities are stale