product-os: Command for Claude Code

Weekly Review

End-of-week review: assess progress against goals, detect stale work, evaluate how you operated, check in on how you're feeling, review development priorities and AI/automation usage, identify actions, and set goals for the next period.

Instructions

Step 0: Parse arguments

• The user provides arguments in the format: $ARGUMENTS
• Expected format: {initials} (e.g. CR or JK)
• If no arguments are provided, check the auto-memory (MEMORY.md) for the user's initials before asking
• Use initials for all file paths (e.g. team/{initials}/todo_list.md, team/{initials}/weekly-reviews/)
• From the auto-memory or team context, identify:
  • The PM's name and role/level (e.g. Product Director, Senior PM, Lead PM)
  • Their development priorities if known (from performance reviews, coaching notes, etc.)
  • Their team/squad and the outcome metric they own (from context/current-priorities.md)

Step 1: Gather data

Pull from all available sources. Run independent queries in parallel.

1. Previous weekly review — Check team/{initials}/weekly-reviews/ for the most recent review file. Read it to get:
   • Last week's goal(s) — this is what you'll assess progress against
   • Themes or issues flagged last week — check if they recurred
   • Any monthly/quarterly goals if previously set
   • Previous development priority ratings and AI/automation observations — check for trends
2. Todo list — Read team/{initials}/todo_list.md and identify:
   • Items marked as done ([x]) — this week's completions
   • Items marked as dropped ([-]) — removed because no longer needed
   • Items still open — note overdue or expected-this-week items
3. Git history — Run git log --since="7 days ago" --oneline --stat to see what was committed this week
4. Recently modified files — Run git diff --stat HEAD@{7.days.ago}..HEAD to understand scope of changes
5. Remote sync check — Run git fetch then git status to check if the current branch is behind its remote tracking branch. If there are remote changes not reflected locally, flag this to the PM before proceeding
6. Meetings — Use mcp__claude_ai_Granola__list_meetings with this_week and last_week to get all meetings from the past 7 days. Then use mcp__claude_ai_Granola__get_meetings (batch, up to 10 IDs per call) to pull notes and summaries from each. If a meeting's notes are sparse (just a few words or empty), use mcp__claude_ai_Granola__get_meeting_transcript to pull the full transcript and summarise key points instead
7. Current priorities — Read context/current-priorities.md to understand team goals and Q-level priorities this PM is responsible for
8. Goals — Read team/{initials}/me/goals.md for monthly, quarterly, and development priorities
9. Achievements — Read team/{initials}/me/achievements.md for current state
10. Decision log — Read team/{initials}/decision-log.md (if it exists) for this week's decisions and unresolved follow-ups
11. Recurring meeting context — Read files in team/{initials}/recurring-meeting-context/ for carry-forward item trends
12. Claude Code usage — Read the auto-memory files. List the commands in .claude/commands/. Review git history for patterns in what Claude was used for (command runs, file types created/edited, research vs execution). Check .claude/commands/ and scripts/ for new or modified automations
13. Quarterly tracking changes — Check for team/{initials}/quarterly-tracking-changes-*.md files from this week. If found, read them and incorporate any RAG changes, new risks, or milestone updates into the Progress Review