Quarterly OKR ritual + auto-status collection

A quarterly OKR cadence where the status column is collected automatically instead of nagged out of people. An agent reads task trackers, Git activity, chat threads, and recent docs, and proposes a status update per OKR with evidence links. Owners review and edit rather than write from scratch.

The execution gap between strategy and reality lives in the "I'll update the status doc on Friday" lie. Auto-collected status closes the gap from ~40–60% to ~80–90%. At a large wealth-management firm this pattern is approved as a board-level cadence, with quarterly priority streams per business line.

1. Pick the cadence (quarterly works; monthly burns out reviewers).
2. Define each OKR as a markdown file with: objective, key results, owner, evidence sources (task tracker filter, repo path, chat tags).
3. Skill okr-status runs weekly: for each OKR, pull the linked evidence sources, summarize progress, propose a status (on-track / at-risk / off-track) with citations.
4. Owners review the proposed status, edit, commit. The full history of status updates lives in Git.
5. Quarterly review meeting reads the latest auto-collected status — discussion focuses on what to change, not "what's the actual number."
6. At quarter end the agent drafts the next quarter's OKR list from the previous quarter's misses + current strategy doc.

• Status that's auto-collected but never read becomes wallpaper. Tie it to a real meeting with real decisions.
• Don't let the agent set status without owner sign-off — only propose. Otherwise everyone learns to ignore "on-track" since the agent never escalates.
• At early-stage scale (~30 ppl) OKRs may be premature; lighter weekly auto-collected priorities work better.