# Steal or Open

> A house-rules check for holiday gift swaps of the steal-and-unwrap kind: everyone brings one wrapped gift, and on a turn a player unwraps a new gift or steals an opened one. Paste your rules, one per line, and the page reads them, simulates hundreds of games on a trading-floor board, and - in one metered run - reviews them like a game design document: lines that clash, loopholes walked step by step, what is missing, a verdict, a clean table card and a host script.

URL: https://steal-or-open.skillsafe.ai/
API tutorial: https://steal-or-open.skillsafe.ai/api.html

## Free, in the browser
- Numbering the lines (list markers stripped; at most 30 lines of 280 characters; a .txt or .md file can be dropped in).
- The topic reading, from word lists: the five core topics every swap needs - what a turn is, turn order, price cap, steal limit, how it ends - and five extras - stealing straight back, steals per turn, wrapping, showing opened gifts, after the game. Each topic shows the lines that seem to mention it.
- What the simulator reads from the lines: the steal limit (a gift freezes after so many steals; two different limits are flagged), whether stealing straight back is barred, a per-turn steal cap, a first-player final turn, and a price.
- The trading floor: 400 seeded games with your player count and a crowd that steals 25%, 45% or 65% of the time when a steal is allowed. It shows the estimated game length (60 seconds per unwrap and 25 per steal are assumptions, not timings), steals per game, the longest single turn, how many gifts get frozen, and one sample game as a board of gift tickers (each steal a tick up, frozen gifts HALTED) and a ticker tape of moves.
- Exports: Markdown, JSON, a printable table card, "Share with guests" (the card as a plain message for the group chat or invite), copy the card or the host script; a re-review says how it compares with the last one; "Use the card as my rules" replaces the rules with the card so a second review checks the patched set; kept rule sets (in the browser, or in your account when signed in).

## Metered (gpt-terra), one task: rules-check
The verdict follows a fixed check order, first match wins:
1. rewrite - two or more lines clash, or no line says what a player does on a turn.
2. patch - a core topic is missing (turn order, price cap, steal limit, how it ends), or there is at least one loophole.
3. ready - otherwise; missing extras are allowed.
The reply lists clashes (the lines involved, why, a fix), loopholes (a step-by-step situation the rules do not settle, a fix), gaps (one per missing topic, a fix), a table card of at most 10 rules in play order with every fix marked as suggested, and a host script of at most 120 words. The model works only from the lines and the page's figures: no gift ideas, names, genders, prices or numbers the rules do not hold (only a suggested fix may propose a number). The page re-checks every reply: cited lines real, the verdict matching the reply's own lists, the card keeping the rules' price and steal limit, unmarked card rules citing lines, and no line id in prose.

## Sources
Built on the SkillSafe skill @donchitos/design-review - https://skillsafe.ai/skill/@donchitos/design-review - (review a game design document for completeness, internal consistency and implementability before it goes to the people who build it), re-aimed at the house rules of a holiday gift swap.
The trend note on the page (readers of the English Wikipedia article on the steal-and-unwrap gift exchange up about 65% in the week to October 20, 2026, from about 133 to about 219 views a day on a 7-day average) is from the Wikimedia pageviews API as fetched on October 21, 2026.
The simulator's per-action times and steal chances are the page's own assumptions; every minute figure is an estimate.