Every change Wellknown observed on this MCP server, newest first, with what it was before and what it became. Tool-surface changes carry the definition diff. Nothing here is edited after the fact.
Certificate changed, valid to 2027-01-05
Changed the definition of "find_firms"
⟨150 unchanged words⟩ description":"Minimum trading days at or belowthis.this, counted as minTradingDaysUnit names. 16 of 71 record none.","type":" ⟨54 unchanged words⟩ winning_days. A wait is compared only with waits counted the same way; a firm counting another way, or stating no unit, is returned under unknown with what it counts named.","enum":["calendar_days","trading_days","winning_days"],"type":"string"},"minTradingDaysMinProfit":{"description":"With winning_days only: the profit a day must reach to count. A firm whose winning days carry a different profit bar is returned under unknown.","type":"number"},"minTradingDaysUnit":{"description":"What maxMinTradingDays counts: calendar_days, trading_days or winning_days. A minimum is compared only with minimums counted the same way; a firm counting another ⟨22 unchanged words⟩
Changed the definition of "find_firms"
⟨129 unchanged words⟩ Minimum days to first payout at or belowthis.this, counted as minDaysToPayoutUnit names. 26 of 71 record none.","type":" ⟨27 unchanged words⟩ 71 record no fee.","type":"number"},"minDaysToPayoutMinProfit":{"description":"With winning_days only: the profit a day must reach to count. A firm whose winning days carry a different profit bar is returned under unknown.","type":"number"},"minDaysToPayoutUnit":{"description":"What maxMinDaysToPayout counts: calendar_days, trading_days or winning_days. A wait is compared only with waits counted the same way; a firm counting another way, or stating no unit, is returned under unknown with what it counts named.","enum":["calendar_days","trading_days","winning_days"],"type":"string"}},"type":"object"}
Changed the definition of "find_firms"
⟨24 unchanged words⟩ ,"crypto"],"type":"string"},"dailyLossLimitUnit":{"description":"The unit maxDailyLossLimit is in. Daily loss limits are recorded in percent at some firms and in dollars at others, and the two are never compared: a firm recorded in the other unit is returned under unknown with its unit named.","enum":["percent","usd"],"type":"string"},"maxActivationFee":{"description":"Activation fee at or ⟨26 unchanged words⟩ description":"Daily loss limit at or belowthis.this, in the unit dailyLossLimitUnit names. 31 of 71 record none.","type": ⟨53 unchanged words⟩
Added "compare_firms", "find_firms", "get_evidence" and 1 more (2 tools before, 6 now)
Two firms side by side. Reports each figure with its evidence state and whether the pair is comparable at all: a field neither firm records is not a tie, and a field only one records is not a win. No per-field winner is declared, because the direction of "better" for several of these is a judgement the published scoring model already makes; each firm's grade is returned alongside so the verdict comes from that model rather than from this tool.
Firms matching every criterion given. A firm whose record does not carry a criterion is never counted as a match and never silently dropped: it is returned separately under unknown, with the field that could not be tested. Absence is common here, so read counts before treating the matched list as the answer.
The stored reading behind a figure: the verbatim quote, the page it came from, and the date we read it. Answers "we hold none" explicitly rather than returning an empty list. `outcome` is load-bearing: a `contradicted`, `contradiction` or `absent` row records what we read, not a figure we publish. Use get_firm for the published figure.
Certificate recorded, valid to 2026-11-07
Authorization not required
First tool surface recorded: 2 tools (server version 1.0.0)
Showing the latest 8 events. The API returns up to 500 and filters by kind: ?kind=tool_surface_changed
The dated record of what changed, newest first. Every entry states its KIND, and most of this tape is FirmFax correcting its own data rather than a firm changing its rules. Read counts before describing industry activity: a total taken across kinds overstates it.