# Wording synthesis — 14 September 2026

All four requested reviewers assessed the same frozen source. Reports:
[Claude Fable](fable.md) (`claude-fable-5-1`), [Gemini via agy](gemini.txt)
(`gemini-3.1-pro-high`), [Zai](zai.md) (`glm-5.3`), and
[fresh Codex Astra](astra.md) (`gpt-6-astra`). Fable and Zai had no tool permission
denials. Gemini's first file-reading attempt was denied by headless agy; the
successful second attempt received the source inline and used no file tools.
The [brief](brief.md), frozen sources and hashes preserve the reviewed version.

## Decisions

| Finding | Decision | Result / reason |
|---|---|---|
| Intent missing from long/short | Adopt all four | Bullish and range-bound intent appears before target trading behavior. |
| Rename Covered short to Covered call | Keep current name; add intent | Maintains the user's long/short vocabulary without implying a conventional option contract. |
| Replication sounds like a promised payoff | Adopt Fable/Astra; strengthen Zai/Gemini suggestions | Use “targets long-call-like exposure” and “target holdings.” Do not retain unqualified “replicates.” |
| Short spread / premium relationship absent | Adapt all four | State the user's target to cover option premium over time. Explicitly state that no separate option premium is paid or received at entry. |
| “Earns its option premium” / profitable round trips | Reject stronger language | A target spread is not proof of realized premium or profit. |
| Collateral always borrowable | Adopt all four | State Aave reserve and collateral conditions; retain current read-only scope. |
| Current assets confused with original deposit | Adopt Gemini | “Current supplied assets” preserves custody meaning and clarifies timing. |
| Net account value mistaken for profit or exit | Adapt all four | Keep transparent “Collateral − debt” arithmetic and place a compact qualifier beside the figures. |
| Rename net figure to Net Aave value / balance | Reject | Adds a new term without explaining the subtraction. |
| Pending order / fill-or-kill wording | Adopt Astra/Fable | Previewed trade, not pending trade; full fill required, without implying immediate cancellation. |
| “Orders the contract would sign right now” | Reject | Too technical and too easy to read as posted orders; validity is tied to the displayed snapshot. |
| Health restriction above 1 | Adopt all four | Explain the outgoing transfer before incoming proceeds; distinguish liquidation at 1 from the example's 1.001 trade check. |
| Large hedge changes two variables | Adopt Astra/Zai | “More debt + larger trade” names both changes. |
| Address/sample banners imply identity/disconnection | Adopt Astra | Describe the viewed data without asserting that a wallet is disconnected or cannot match the target. |
| Builder inherits portfolio banner | Adopt Astra | Builder and saved examples have their own route-specific scope text. |
| Order absence list misses possible causes | Simplify | State unavailable preview without diagnosing a cause that was not read. |
| Missing strike / expiry explanation | Adopt Fable | Add reference-price meaning and the five-minute terminal trading window, without promising automatic payout. |
| “Factory verified” sounds like safety endorsement | Adapt Astra | “Base trial factory,” with creation verification in engineering notes. |
| Long health-after field mapping questioned | Verify; no change | `positions.js` passes `hfAfterAsk`/`hfAfterBid` into each actual order's `healthAfter`. No mapping defect was found. |
| Missing order-floor input in builder | Keep preview scope | The builder saves examples; opening requires additional settings and a funding quote. No new transaction control was added. |
| Futures / perpetuals analogy | Defer | No unsupported product mapping, funding rate or perpetual term was introduced. |

Canonical strategy copy now lives in `model.js` and is reused by landing,
builder and live-card descriptions. The Astra recheck cleared the high-impact
fixes and requested an entry-premium clarification on the short summary; that
clarification was added directly to the shared note.

Product authority: the user's stated economic intent plus current
`thetaAaveCattle/src/ThetaCattlePair.sol`, lens and frontend notes. Public Aave
reference checks: [supplying assets](https://aave.com/help/supplying/supply-tokens)
and [Aave V3 borrowing and variable rates](https://aave.com/help/aave-101/introduction-to-aave).
These sources explain Aave; they do not define Cattle's spread objective.

## Later user steering

After the wording pass, the user requested order presentation that distinguishes
the short's two resting limits (buy below, sell above) from the long's single
market rebalance toward its target. This is a separate follow-on visual pass,
recorded sequentially in `designs.html`. Long thresholds must not be invented from
the short's grid or the option strike. The current contract emits one bounded
marketable order for the long's eligible target gap, not two resting limits.

## Verification limits

The 60 existing tests pass after the copy changes. Browser checks cover the
rendered wording, routes, dialogs and theme. Four model reviews are complementary
reasoning, not a human comprehension study. No wallet transaction was submitted.