Day 22: Set alerts on score moves

When does a score change matter? When commit velocity flips by ≥0.4.

Yesterday

Yesterday you documented the system. Today you add the layer that catches the move between weekly checkpoints.

Why this signal matters

A weekly snapshot is a habit. An alert is leverage. The right alert threshold catches the score move before the round closes; the wrong threshold drowns you in noise. Calibrating thresholds is the difference between a watchlist and a deal-flow engine.

The 5-minute exercise

1Set a Slack/email alert (or just a Calendar reminder) on commit velocity flips ≥0.4 week-over-week.
2Add a second alert on README updates that change >50 lines.
3Add a third alert on new repo creation (any new repo in a watched org).
4Test by triggering one manually on a known org.

What you’re filtering for

Three alert types live, each with a clear noise floor. If you get more than 3 alerts/week per org, the threshold is too sensitive — narrow it.

Edge case

GitHub doesn't natively send velocity-change alerts. The MCP server has the tool; the manual workaround is GitHub's own watch settings + a weekly Monday recheck. Pick one.

Bonus

When an alert fires on a watchlist org, the response should be tight: 60-second context check, 5-minute composite recheck, a meeting request if the score crossed a meaningful threshold. Document the reaction loop.

Tomorrow

Tomorrow: the anti-signal — the orgs where 6/6 is wrong, and how to flag those before you waste a meeting.

Curriculum: /challenge · Methodology: /methodology · Paper: ssrn.com/abstract=6606558