# What Was That?

> Sort an odd light or object you saw in the sky into everyday suspects. Free in the browser: pick up to 8 details and see 9 suspects scored, ranked and plotted on a radar scope. One metered run explains the top 3 both ways and suggests what to watch for next time.

## What it does
- The form asks 8 details, each of which may be "not sure": when (daytime, dusk, night, before dawn), how long, the lights (a single light, 2 to 5, a line of 6 or more, a fixed shape), colour, motion, sound, how it ended, and height. Optional free notes are passed to the run but not scored.
- The page scores 9 everyday suspects: an aircraft, a single satellite or space station, a string of newly launched satellites, a bright planet or star, sky lanterns, a drone, a meteor or fireball, a high-altitude balloon, and a reflection on glass. Each detail fits a suspect, rules it out, or leaves it open.
- Ranking: fewest details against, then most details that fit, then a fixed order (aircraft first, reflection last).
- Verdict rule, in order: still open (even the top suspect has a detail against it); toss-up (the runner-up has nothing against it and fits at most 1 detail fewer); solved (the top suspect fits 6 or more details); likely (otherwise).
- The radar scope puts each suspect at its own bearing, nearer the centre the more details fit; ruled-out suspects sit on the rim.

## The paid task
One task, `sort-sighting`, on gpt-terra, metered per run. Input: `task`, `sighting` (8 lines), `notes`, `fit_table` (9 ranked lines). Reply: one JSON object with `verdict`, `headline`, `suspects` (the top 3, each with `id`, `fits`, `against`, `test`), `note` and `closing`. The page re-checks the reply against its own rule.

## Exports
Markdown, CSV fit table, JSON, the radar as SVG or PNG, print.

## Notes
A rule of thumb for curiosity, not an official finding. Never look at or near the sun; never shine a light or laser at an aircraft; sky lanterns are a fire risk.

## Sources
Built on the SkillSafe skill @lilmgenius/factchk (https://skillsafe.ai/skill/@lilmgenius/factchk).
- App: https://what-was-that.skillsafe.ai/
- API: https://what-was-that.skillsafe.ai/api.html