Don't trust the bar. Replay it.

What is checked, coin by coin

• Same trades, same boss. Your browser replays the trade list twice and gets the same fallen segments, at the same second, for the same reason.
• Site and browser agree. The fallen segments computed here match the ones the site reports.
• Every burn is a real transaction. On mainnet, each burned slice links to its signature on Solscan. A slice that could not burn is shown in red with its reason.
• No bundle. The creation transaction holds the coin and the dev buy, nothing else.

What we tested before going live

• A second, independent implementation of the rules, written second by second, gives the same boss on simulated histories and on hand-written edge cases.
• Every test has a sabotage mode (a brake of 599 seconds instead of 600, one token missing, a forgotten trade): each one must fail, and does.
• The trade reader was run on real Stonks curves: buys minus sells equal the tokens the curve sold, to the token, and pool accounts are never counted as holders.
• The 12 pre-signed burns were sent on a copy of mainnet minutes after signing: each burned the exact slice, in order, and a slice the dev's bag no longer covered was refused by the chain.