# For Services To

> A nomination desk for small awards that do not exist yet. Invent an award for someone who quietly does a good thing - a colleague who keeps the tea shelf going, a flatmate who takes the bins out, a neighbour who fixes every bike on the street - give it two to four criteria, list what they actually did, and see every line pinned to a prize rosette. One metered run reads the evidence against the criteria and drafts a citation, a nomination and three questions for the nominator. https://for-services-to.skillsafe.ai/

## What it does

- Free, in the browser: reads the award's name, the nominee, what the award is for, 2 to 4 criteria (one per line) and 3 to 12 evidence lines (one thing the person did per line). It numbers the evidence, marks lines that carry a number or a time together with an action, and lines that read like praise only, guesses which criterion a line fits by shared words, and draws a prize rosette with one ribbon tail per criterion and one band per line. Exports: Markdown, CSV, SVG (the rosette) and JSON, plus copy or share of the drafted nomination.
- Metered, one task (`nominate`): every evidence line is read as specific or vague and filed under up to 2 criteria; one verdict by a fixed order - rename the award (fewer than half of the lines fit any criterion), gap in the evidence (a criterion has no line), too vague (more than half of the fitting lines are praise without an act), otherwise shortlist ready; a better name when renaming; a citation that starts "For services to"; a nomination of at most 110 words in the nominator's voice; three questions for the nominator, the first about what decided the verdict. The page re-checks every reply against the form. Each of the panel's questions has a button that puts the caret on the line it is about (or on a new line) in the evidence box, and a suggested better name can be applied in one click; nothing runs until the person asks again.

## Contract

- Input (all strings): task, award, for_what, nominee, criteria ("c | text" lines), evidence ("n | text" lines), page_facts, optional clip_note and retry_note.
- Output: one JSON object with verdict, verdict_reason, lines [{n, criteria, kind, shows}], criteria [{c, reading}], better_name, citation, nomination, questions [{line, criterion, ask}], closing.
- API tutorial: https://for-services-to.skillsafe.ai/api.html

## Notes

- A playful way to say thank you, not a real awards scheme. Ask before you share anything personal about someone.
- Nothing is kept unless the person presses save; saved nominations go to their own account.
- Source skill: https://skillsafe.ai/skill/@jezweb/award-application (award submissions that map achievements to selection criteria with evidence).