Check the oracle feed surface against SEP-40

[luchobonatti]

Research question

Does the nav-oracle's read surface conform to SEP-40, and is the delta small enough to adopt the interface?

Time-box

1 day

Background / What we already know

The feed exposes nav_per_share, latest and state, designed against the Chainlink NAVLink field set rather than SEP-40, the Stellar price-feed standard. Conformance was never checked, so an integrator expecting the standard interface finds something adjacent to it.

A clean fit is not guaranteed. nav_date is the off-chain valuation date, distinct from ledger time, and expires_at is a hard staleness boundary. Neither has an obvious SEP-40 counterpart.

Investigation approach

1. Map the oracle's read surface field by field against SEP-40
2. Note what matches, what is named differently, what is missing on each side
3. Check whether a conformance shim can preserve the consumable gate
4. Decide: adopt, or record the divergence and defer

Expected output

• the field-by-field delta written down
• a decision, either way
• if adopted: the interface implemented, with a conformance test
• if deferred: the divergence recorded in this issue with its reason, and carried into the runbook for integrators
• the decision recorded as an ADR either way