[argszero]

Summary

The README's operator-overview section describes the trades card as "the 9th card". It is not. The #ops-stats grid that renderOps() renders is built by the array at ui/js/app.js:3173-3189, which emits exactly nine stat(...) calls in this order:

#	key	card
1	ops.stats.status	运行状态
2	ops.stats.version	服务版本
3	ops.stats.uptime	运行时长
4	ops.stats.users	用户数
5	ops.stats.keys	上架 key 数
6	ops.stats.trades	交易量
7	ops.stats.calls	本月调用
8	ops.stats.in	点数流入
9	ops.stats.out	点数流出

The trades card is the 6th. What the words mean here is fixed by the repository's own usage: the sibling comment at ui/js/app.js:1132 says "the 4th of the prototype's 4 cards" for the fourth card of #share-stats, so 「第 N 张」 = position in the card group, and the same README sentence already counts the group ("the prototype has 4, this has 9").

Related Issue

None — the in-repo variant of a documentation claim that had drifted from the code.

Changes

• Drop the ordinal from the 交易量卡（第 9 张，回归修复） bullet, leaving 交易量卡（回归修复）. The rest of the sentence is untouched.

The ordinal is dropped rather than refreshed. Nothing owns the grid's order — no gate and no constant pins the sequence of stat(...) calls — so a corrected number would rot the same way on the next card that gets inserted, exactly as this one did. The "prototype 4 / implementation 9" count in the same sentence is left alone: it is a group size, not a position, and it is correct (verified against docs/prototype/aitokenpool-console.html:1385, which renders four statCard(...) calls).

Because the card was placed at that slot deliberately — the landing note records it as "immediately after the keys card, i.e. the mock branch's original position" — the position itself is not a defect; only the number written about it was wrong.

• No config or data-structure change

Tests

• cargo test — 557 passed / 0 failed
• cargo fmt --check — clean
• cargo clippy --all-targets -- -D warnings — clean
• No new unit test (documentation-only change; ui/README.md is part of the include_str! corpus, so the full suite was run against the edited tree)

Checklist

• Branch name follows the convention (docs/)
• Commit message uses Conventional Commits
• Single responsibility, minimal change (1 file, +1/−1)

[argszero]

Self-review note.

Claim being removed. 交易量卡（第 9 张，回归修复） → the trades card is the 6th of the nine stat(...) entries the render array emits (ui/js/app.js:3173-3189): status, version, uptime, users, keys, trades, calls, in, out.

What fixed the reading of 「第 N 张」. ui/js/app.js:1132 calls the fourth card of #share-stats "the 4th of the prototype's 4 cards" — an ordinal over the card group — and the same README sentence under review counts that group ("the prototype has 4, this has 9"). So 「第 9 张」 claims a position, and the position is 6.

Why drop rather than correct. Nothing owns the grid's order: no gate and no constant pins the sequence of stat(...) calls, and git log -S shows the number was authored in the same commit that inserted the card (4830248), whose own note says the card goes "immediately after the keys card, i.e. the mock branch's original position" — position 6 was the intent. A refreshed number would rot on the next insertion the same way.

Scope check. ui/README.md is include_str!'d, so the edited tree is the one the suite measured: cargo test 557/0, fmt --check rc=0, clippy --all-targets -- -D warnings rc=0. Diff is 1 file, +1/−1.