What shipped
Kill Criteria is a self-contained browser tool for deciding in advance when a bet should stop. Title the project, write a one-sentence purpose, name an owner, pick a review-by date, and choose a default action if criteria trip (Pause / Pivot / Kill / Scale down). Add measurable criteria — metric, operator, threshold, unit — then mark each as Unknown / On track / At risk / Tripped. A live signal panel shows DRAFT / ALIVE / WATCH / TRIPPED / OVERDUE from those statuses and the review clock. The markdown receipt updates live; Copy receipt dumps it for Slack, Notion, or a steering meeting. Presets cover AI ops pilot, Feature bet, and Vendor eval. Optional localStorage under kill-criteria-v1. Heuristic only — status chips are manual judgment, not live telemetry.
How to try
- Open the live demo (iframe on the right, or full-page link below).
- Load a preset (start with AI ops pilot).
- Flip a criterion to At risk or Tripped — watch the badge swing to WATCH / TRIPPED.
- Hit Copy receipt and paste into a steering channel or agent prompt.
Scope trim
No live metric connectors, no OKR sync, no calendar invites, and no multi-stakeholder voting. One bet, a handful of criteria, one heuristic signal, one markdown dump, local persistence only. The point is to write the stop rules before sunk cost takes over.
Limitations
- Statuses are manual — nothing polls your analytics or tickets.
- Clipboard copy needs permission; otherwise select the receipt manually.
- Timestamps use the browser’s America/Chicago-friendly locale formatting when available.
- Not a formal portfolio or stage-gate system — use it to think clearly, then follow your real process.
Suggested next steps
- Optional “evidence link” field per criterion for dashboards or tickets.
- Export a .md download in addition to clipboard copy.
- Tiny library of org-specific criterion templates beyond the three presets.