Nine research and data reports plus a synthesis, built from Thomas's two voice notes of 00:08 and 00:28. Everything before tonight is experimentation. Start with the synthesis, then the wireframes on a phone.
Synthesis: what the night established, contradictions, module list, questions
Sources and rights
Infrastructure and architecture
Layer 1: follow the money
Discovery intake and attention
Situation model: fields and lights
Zones and the forgiving win
Presentation and competitors
Wireframes v0
Beta and commercial
The rendered track after the options: d2 Tracks carried through twelve passes, the frozen Now built on the real read model, and the contract that feeds it. Newest first. What needs Thomas. The road to the live feed (milestones, pieces, owners).
The coverage print: an interval attested by Rook's span sidecar through Bolo's adapter prints its measured value with the attestation in its title, everything else stays unread. 哈基米 re-assembled at 07:27Z on 29 Sep from its 119 covered BSC ranges, placed on the clock by two block header reads (06:00:27Z to 07:23:10Z) and reconciled 146 tape rows against 146 logs; 2 intervals print a measured $0 in, $0 out, 0 swaps with the attestation, 13 inside the union keep usd conversion unavailable, 3,062 stay unread; a Money in/out strip and a coverage line on Board, Now and Situation; check-b11 0 placement disagreements against Bolo's records; 螃蟹 (0 attested, every interval inside its union held swaps with no USD) and STONK (no span file) stay build-10's bytes; MC $34.72M as of 28 Sep 23:15Z under the unchanged cap rule; 1,142 checks pass; no contract change
Three BSC caps on a read basis with no provider at all: GENIUS and TVS (quoted in USDT) and HPA (quoted in USD1), priced on their own quote under the contract's tape_usd rule (a USD stable quote is its own USD value, no conversion, no mark). Each quote bound by address before assembly: USDT 0x55d3...7955 on every USDT pool the tape names, USD1 0x8d0d...8b0d, all three agree with the tape's pool record. Fresh totalSupply reads and tape pins at 06:55Z on 29 Sep; MC $318.07M as of 06:54Z, MC $519K as of 06:53Z and MC $10.38M as of 06:55Z on Market, Board, Now and Situation, equal to Python to the cent; no CoinGecko call, no contract change; one page edit, the cap clock's title names the quote on our tape; every flow interval unread
The BSC USD path on a read basis: route A asked CoinGecko for the quote token itself and got 404, coin not found (pinned), so 螃蟹 stays build-9's bytes and still cap not read; route B held one BNB/USD range at 23:20Z and chose the richest WBNB pool on our tape trading within the hour (哈基米, 37,245 rows, 15 of 39 BSC tapes qualified), assembled at 23:23Z from a fresh totalSupply read with the base bound by its last Swap log; MC $34.40M prints with as of 23:15Z and Powered by CoinGecko on Market, Board, Now and Situation, equal to Python to the cent; every flow interval unread; no page changed
The Money in/out track directly under Wallets in on Situation, from Thomas's 14:30 line; Now, Board and Market byte-equal
Cap leads, no estimated caps, the three labels, Board frozen
Both real documents under the coverage rule, checked against Bolo's B10 slices: the BSC document re-assembled at 20:25Z from a fresh supply read and tape pin, its 105 intervals that carried 0 with no attestation now unread with the reason source coverage unknown, all 1,757 intervals null with a reason and none a bare zero; every interval up to the build-7 cutoffs agrees with Bolo's B10 reasons on both chains, 0 disagreements (BSC 105 plus 1,568, STONK 3,338 plus 36); STONK is build-8's bytes; no page changed, BSC still cap not read; CHECKS.md and the six BSC shots at 1280 and 390 are on the build page
A live conversion at assembly time: one CoinGecko SOL/USD range held at 20:27Z with its own provenance, the pool chosen from our tape (ALLINU on meteora, the richest of 51 Solana tapes read within an hour of a point), a fresh supply read and tape pin; MC $4.97M prints with as of 19:04Z and Powered by CoinGecko on Market, Board, Now and Situation, equal to Python to the cent; no page changed; STONK and BSC print exactly as in build-8, BSC still cap not read
The STONK cap on a read basis: the Solana document re-assembled at 18:47Z from a fresh supply read, a fresh tape pin and one pinned CoinGecko SOL/USD reading that covers the 44 reads in the hour before 10:00:18Z on 26 Sep; MC $34.32M prints with as of 26 Sep 10:00Z and Powered by CoinGecko on Market, Board, Now and Situation, equal to Python to the cent; every STONK flow interval unread, never a zero we did not measure; BSC byte-equal to build-7, still cap not read
The BSC supply admitted under contract v3.7: 螃蟹 re-assembled from the pinned bytes with its 18-decimal supply read (1,000,000,000 tokens), Board and Market print the supply line, and the cap slot stays cap not read titled no USD price at this moment because the pool has no USD price; the wide fixture prints MC $61.8K from an 18-decimal supply, equal to Python to the cent; Now and Situation unchanged
Now stops printing 0 entered on a null cumulative: contract v3.6 applies §7.4 (partial coverage never prints as zero) in nowat.py and nowat.js with parity, the slot reads entered not counted with the document's reason as its title on both real rows, Now byte-equal to build-5 on the fixtures, Board, Market and Situation byte-equal on every document, and the six Solana shots
Board and Market on the real BSC and Solana documents, the card and the eight lines built from the row alone with every USD slot refusing in the contract's words; the Now zero-maximum bars and the Situation price cell fixed, 0 console errors on both rows
The Money in/out track on the real documents: pass 13 read from rows[].flow under v3.5, and the first real tokens (螃蟹 on BSC, STONK on Solana) on Now and Situation with no USD figure printed anywhere
The pass 12 strings and the cap rule on the real Solana document: a figure only on a read supply, else cap not read; a synthetic supply fixture prints MC $2.63M
The two pass 10 phone defects, nothing else
Phone forms of Now, Situation, Board, Market
Market
Board
Now, frozen
The frozen Now on Bolo's real Solana document: the first prospective history, zero rows with the refusal printed
Now and Situation re-based on pass 11, on the v3.4 documents; dead names refused on the page
Board on the v3.4 read model, pass 11 strings byte-equal, entry cap from the document
Frozen Now on the real SPRING envelope
A USD conversion covers a tape read only within 60 minutes of its clock, counted from the moment we held it; a printed cap carries its clock and the provider's mark. v3.9 takes the conversion product exactly in Python as in JS; v3.10 prints cap not read where a USD price has no supply; v3.11 makes the flow totals and Board's entry cap exact, with the v3.10 null rule on the entry cap. Before: v3.7 widened a supply read to 78 digits
Two rendered directions inside the decided envelope, on the real SPRING numbers. A, Instrument: number first, cool, ledger-dense. B, Survey: sentence first, warm mineral, field-book margin. Open the options page or read OPTIONS.md.
Hierarchy fix after the external V2 review: state word and one sentence first, the clock as the spine, since you looked and waiting for above the fold, evidence below, data notes as a drawer. Tabs Now, Situation, Board. No lights row on the list, no profile labels, no zones page. CHANGES.md maps every move to the review.
Black on white, drawn slots, illustrative values, working links between pages. No visual design. v0 grey boxes still available.
Mandate · Index of every artifact, script and CSV · Filed at projects/caverio/companion-2026-09-26/