bStocks: Binance-sourced synthetic markets in /v2/rates - #39
Merged
Conversation
Adds a Binance singleton provider (mirroring CoinGecko's shape) that fetches the tokenised bStocks universe and Spot 24h/7d tickers from Binance's public endpoints, no API key required. Exposes getTokenisedAssets (1h cache, BSC-only filter), getTradingSymbols (1h cache, TRADING-status filter from exchangeInfo), getTicker24h (60s cache, single batched call), and getTicker7d (60s cache, chunked to 20 symbols/request via the existing arraySplit util). Ticker fetches keep a last-known-good value per symbol so a halted/omitted symbol (e.g. during a stock-split trading break) or an outright failed refresh still serves the previous price instead of dropping the entry. Fixtures are built from live curls of both Binance endpoints rather than assumed shapes. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
Binance does not omit a halted symbol from the ticker response - it returns the symbol present with lastPrice "0.00000000". Measured live: of 20 BREAK-status symbols requested, 20 came back present and 9 were priced at zero. So the absent-symbol fallback was dead code for the exact scenario it was written for, and the zero was accepted as fresh AND written to the last-known-good store - poisoning it, so a later outage would serve $0 forever rather than the real last price. mergeTickers now admits only finite, strictly positive prices, both to the served set and to the store. Adds lastGoodAgeMs() so a caller can tell a live price from one carried through a long halt, and sorts the ticker cache keys so the same request in a different symbol order still hits cache. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
Intersects the Binance tokenised-asset universe (BSC-listed, from Task 1) with TRADING Spot symbols to emit CryptoPrice[] entries for bStocks, quoted against BTCUSDT fetched in the same batch so both legs share one venue. Ids are bstock-<code> under provider "coingecko" per the cross-repo id contract with the sibling api repo. A module-level last-known-good map keeps serving a symbol's previous price across a refresh where it's omitted (CEX halt around a stock split), rather than dropping it. Verified against live Binance endpoints: 66 tokenised assets are all BSC-listed, and intersecting with exchangeInfo's TRADING Spot symbols yields 56/66 tradable today (the other 10, e.g. NFLXB/ASMLB, aren't listed on Spot at all yet) — matching the brief's expected ~56/66. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
… path
The guard against a missing or zero BTCUSDT was correct but untested, so a
later simplification could silently put Infinity or NaN into rates.btc with
nothing in CI to catch it - the same shape of bug that reached a signing path
in the sibling api repo. Mutation-tested: dropping the btcUsd half of the
guard fails both new tests.
Also pins that one halted asset does not affect its siblings in the same
batch, and that the feature flag actually suppresses output.
Corrects two JSDoc claims that came from the plan text: the client does no
provider-prefix parsing (applyMarkets and use-fiat.js each build the literal
`${provider}-${id}` and the strings simply have to match), and Binance does
not omit halted symbols - it returns them present at zero.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
Adds a fifth try/catch provider block to zelcoreRatesV2.getAll() that
appends getBstockPrices() (Task 2) into the crypto array, wrapped so a
Binance/bStocks outage sets errors.binance = true instead of throwing out
of getAll(). Also excludes synthetic `bstock-*` ids from the CoinGecko
harvest in coinAggregatorIDs.ts so rates-api never queries CoinGecko for
them.
Fixes a pre-existing positional-merge bug in apiServices.ts: the crypto
refresh merge used mergeDeep, which walks target/source arrays by index.
Since `processed` in zelcoreRatesV2.getAll() is a concatenation of four
independent try/catch provider blocks, its length and per-index identity
shift between refresh cycles whenever any block throws or an upstream API
returns a different row count -- both routine occurrences. Concretely: if
CoinGecko's block (rank/change7d present) fails one cycle while
CryptoCompare's block (no rank/change7d) succeeds, mergeDeep deep-merges
the old CoinGecko entry at index 0 with the new CryptoCompare entry at
index 0 -- id/provider/rates get correctly overwritten, but `rank` survives
from the stale CoinGecko entry, producing a CryptoCompare coin wearing a
foreign coin's rank. Separately, if the new array is shorter than the old
one, mergeDeep's source.forEach never visits the trailing old indices, so
entries missing from the new fetch persist in the output forever with
frozen data instead of being dropped.
Replaces the crypto merge with mergeCryptoByKey, which discards the stale
target and rebuilds the array from source only, keyed by
`${provider}-${id}`. tests/mergeCrypto.spec.ts pins this: one test
documents mergeDeep's corruption (still used unchanged for fiat/rates/
marketsUSD) on realistic shape-mismatched input, one confirms
mergeCryptoByKey fixes it, and one confirms mergeCryptoByKey produces
identical output to mergeDeep for a normal, well-ordered refresh so
existing consumers see no behavioural change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
…merge key The crypto.length > 300 sanity floor was calibrated before bStocks existed. With ~56 synthetic entries added, a degraded CoinGecko response (250 rows instead of ~342, which it returns silently rather than throwing) now clears the floor where it used to be blocked: 250+39+2 = 291 was rejected, 250+39+2+56 = 347 is accepted. Combined with the wholesale replace, that truncates /v2/rates and drops ~100 coins for the cycle. The floor now counts real-provider rows only. mergeCryptoByKey is renamed replaceCryptoByKey and its unused _target parameter dropped. It never merged - it discards target entirely - and the old signature invited a maintainer to wire target back in, which would throw on the first refresh cycle where ratesV2.crypto is genuinely undefined. A mutation run showed a bare spread of source passed every test in the file: nothing pinned that provider is part of the key, that duplicates collapse, or that the last write wins. Two tests now cover those; all three surviving mutants are killed. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
Documents the synthetic Binance bStock entries in /v2/rates: universe selection (BSC contract + TRADING Spot symbol), USDT quoting, the provider:"coingecko" id contract with the client and the sibling api repo, and the halted-symbol/last-known-good behavior. Regenerates the committed docs/ typedoc tree to pick up the new bstocks/binance modules and types. Bumped the typedoc devDependency ^0.26.7 -> ^0.28.0: typedoc-plugin- markdown@4.12 (already the installed version, satisfying the existing ^4.2.7 range) requires typedoc 0.28.x as a peer, so `npx typedoc` - the docs-refresh step this same README documents - was silently broken on a clean install before this change. Dev-only, no runtime effect. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
…ry-forward, outage visibility, refresh bounding Addresses the whole-branch review of bstocks (PR #39): 1. binance.ts: lastGoodTicker was keyed on bare symbol, so getTicker24h and getTicker7d shared one fallback slot. Whichever window last wrote silently overwrote the other's priceChangePercent/quoteVolume, so a 24h fallback could serve the 7-day change (and ~7x-inflated volume) as the 24-hour figure. Now keyed per `${window}:${symbol}`; mergeTickers and lastGoodAgeMs take an explicit window. 2. apiServices.ts: serviceRefresher rebuilt ratesV2.crypto from the fresh fetch alone once the >300-row floor passed. A provider whose block failed this cycle (CryptoCompare's ~39 rows, LiveCoinWatch's 2) would vanish from /v2/rates outright whenever the remaining providers alone still cleared the floor -- new in this branch, since replaceCryptoByKey has no positional stale tail to fall back on. Now carries forward only the rows belonging to providers present in ratesV2Fetched.errors, then key-merges with fresh data last so fresh always wins. 3. bstocks.ts / zelcoreRatesV2.ts: every failure inside getBstockPrices is caught internally, so a total Binance outage looked identical to a healthy refresh while quietly re-serving frozen prices forever. Added isBstocksDegraded() (true when a refresh prices zero symbols fresh while last-known-good is non-empty), wired into errors.binance, and bounded last-known-good to config.bstocksLastGoodMaxAgeMs (7 days, chosen because the outage this exists for -- a stock split halt -- is naturally multi-day). 4. zelcoreRatesV2.ts / binance.ts: a Binance outage cost up to ~92s (AxiosWrapper's retry budget across getTokenisedAssets, getTradingSymbols, and both ticker windows), stalling the refresh of all other providers behind it every 30s cycle. getBstockPrices() is now raced against a 10s timeout, and getTokenisedAssets/ getTradingSymbols negatively-cache a failure for config.binanceFailureCacheMs (60s) so the retry storm doesn't repeat every cycle. The timeout's setTimeout is .unref()'d since Promise.race never cancels the losing branch. 5. config/index.ts: bStocksEnabled now reads process.env.BSTOCKS_ENABLED !== 'false' instead of a literal, so it can be disabled without a redeploy. 6. bstocks.ts: rank is no longer hardcoded to 0 (which sorted every bStock ahead of Bitcoin in ascending rank order) -- omitted, matching the convention CryptoCompare's rows already use elsewhere in this repo. 7. New test coverage: serviceRefresher (previously zero coverage) for the small-provider-outage guard/carry-forward interaction; a total-outage integration case pinning that getBstockPrices never rejects; and a multi-asset property assertion over the coingecko/bstock-<code> cross-repo id contract. Each fix has a test that was verified to fail before the change and pass after (confirmed via targeted git-stash reverts per file). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
Author
Fixes from final whole-branch reviewThis commit (87b3888) addresses all 6 findings from the final review, plus the 3 test-coverage blind spots called out alongside them. 1. 24h/7d last-known-good collision ( 2. Coins vanishing on a small-provider outage ( 3. Invisible total Binance outage + unbounded staleness ( 4. Binance outage slowing the whole refresh cycle ( 5. No env kill switch ( 6. Test blind spots closed: Verification: 🤖 Generated with Claude Code |
…hing The 10s bound added for the slow-refresh finding resolved to [] on timeout, which reintroduced the vanishing-rows failure it was meant to prevent, for bStocks specifically. The provider carry-forward in apiServices keys on the row's provider, but bStock rows carry provider "coingecko" while their failure is reported under errors.binance - so nothing could ever carry them. On a hung Binance the first cycle produced zero bStock rows AND an empty errors object: the wallet shows $0 with no banner. Since AxiosWrapper's worst case is ~23s against a 10s bound, the very first cycle of any outage hits it, and rows then blink out roughly every third cycle as the 60s negative cache expires. The timeout branch now serves the pruned last-known-good snapshot and flags the cycle degraded, since freshPricedLastRun still reflects an earlier run when the race is lost. isBstocksDegraded no longer requires lastGood to be non-empty: a cold start during an outage had nothing cached, so it reported a healthy service that was serving no bStocks at all. It now returns false when the feature is disabled, so "off" stays distinguishable from "broken". Also makes BSTOCKS_ENABLED case-insensitive - BSTOCKS_ENABLED=FALSE silently left the feature on - and documents it in .env.example, where an operator can actually find it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f
mergeTickers backfills a symbol's last-known-good ticker whenever its fresh
value is unusable, and returns a plain BinanceTicker either way, so
getBstockPrices could not tell a live quote from a carried one: any positive
price incremented freshCount and re-stamped lastGood with `at: now`.
Both staleness defenses in the assembler read that timestamp, so both stayed
disabled for as long as the carry lasted:
- isBstocksDegraded() is freshPricedLastRun === 0, so a run that priced
nothing live still reported a healthy service and errors.binance was
never set -- no banner, no signal, indefinitely.
- the bstocksLastGoodMaxAgeMs sweep compares now - at, and `at` moved
forward on every ~30s refresh, so an entry could never reach the 7-day
bound. A symbol that never prices again would be served at its frozen
price forever, which is exactly what that bound exists to prevent.
The reachable trigger is a ticker-endpoint failure while exchangeInfo still
answers: Binance rate-limits the heavy /ticker routes (418/429) long before
exchangeInfo stops responding, and getTradingSymbols' 1h cache keeps the
symbol in `tradable` through an outright outage besides. A BREAK-status halt
is NOT the trigger -- the symbol leaves the TRADING set, so its zeroed ticker
is never consulted and the entry ages out correctly, bar the <=1h the cached
set lags.
Binance now exposes pricedFresh(symbol, window). The last-good store already
records when each symbol last priced, and a batch answered from quoteCache
legitimately carries a price up to one 60s TTL old, so that TTL is the bound
between live and carried -- no new tunable, and it moves with the cache if the
TTL ever changes. getBstockPrices counts only live quotes toward freshCount
and stamps `at` at the time the price was actually live (now - lastGoodAgeMs)
rather than the time it was re-served.
tests/bstocksStaleness.spec.ts drives the real provider with only the HTTP
layer mocked; tests/bstocks.spec.ts mocks the Binance class wholesale and so
cannot see this seam at all. Its two tests were each verified to fail against
a revert of the half they cover and to pass with both halves in place. The
class stub in bstocks.spec.ts gained the two new methods, modelling its raw
fixtures as what they are: batches priced live this round.
Generated docs are left alone. The committed typedoc tree is already stale on
this branch (no pages for getLastGoodBstockPrices or isBstocksDegraded, both
added after cc6bd3d), and regenerating it from the main worktree rewrites all
58 pages with commit-pinned source links -- a separate call from this fix.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AURdZitXfR5N9A28a1b2Hp
`npm run lint` has not been checking any TypeScript in this repo. .eslintrc.js
set `parserOptions.parser: 'babel-eslint'` -- a vue-cli idiom that plain
ESLint ignores, pointing at a package that isn't installed -- so ESLint fell
back to espree and failed on the first type annotation it met:
src/lib/utils.ts
15:31 error Parsing error: Unexpected token :
One parse error per file, no rules ever evaluated. `eslint ./` also only
walks .js by default, so even a working parser would have skipped the whole
src/ tree.
- parser and type-aware config now live in a `**/*.ts` override, so plain
JS (this config, config/*.js) is still linted without needing to be part
of the tsconfig project
- @typescript-eslint 7 + eslint-config-airbnb-typescript, matched to the
repo's TypeScript 5.4; eslint-config-airbnb-base is now a direct
dependency rather than something reached through the React config
- dropped eslint-config-airbnb, eslint-plugin-react and
eslint-plugin-jsx-a11y: React lint plugins in a Node rates API
- lint script is `eslint . --ext .js,.ts --fix`
- the repo's own rules (max-len 300 above all) are applied inside the TS
override too, since an override's `extends` resolves after top-level
`rules` and airbnb's 100-char default would otherwise win
That surfaced 313 problems; 246 were formatting and are auto-fixed here.
Four rules are configured rather than obeyed, each because it argues with a
choice this codebase has already made -- for..of (airbnb's objection is
regenerator-runtime, which an ES2020 Node service never pays),
object-curly-newline's four-property trigger (this repo allows 300-character
lines), prefer-destructuring on assignments (`rates[2] = fetched[2]` does not
read better destructured), and the two import rules that only fire because
config/index.ts and lib/axios.ts deliberately export the same value both
named and default.
The rest are fixed in the code. Two were real defects, not style:
- tests/productionCompareRates.spec.ts declared `const diffs = []` inside
the forEach, shadowing the accumulator it was meant to fill. Every
mismatch was pushed into an array that was discarded, so
`expect(diffs).toEqual([])` could not fail no matter how far local rates
drifted from production. The second block in the same test does it
correctly. Removing the inner declaration makes the assertion mean
something -- that suite needs a local server on :3333 to run, so it is
worth knowing it may now actually fail.
- src/lib/axios.ts returned the setTimeout handle out of a promise
executor, whose return value is unreachable by construction.
Also renamed a `_contract` loop variable that is used (the underscore said
otherwise) and a `cgCoins` local that shadowed the module import. Deliberate
idioms kept their behaviour and gained a targeted disable with the reason:
mergeDeep mutates its target by contract, the provider loops are sequential
to stay inside upstream rate limits, cgTokens is a module-level cache the
refresher reassigns.
Verified: `npx eslint . --ext .js,.ts` clean, `npx tsc --noEmit` clean, and
jest reports the same 41 passing as before this commit. The two
productionCompare suites still fail for the reason they failed already --
nothing is listening on localhost:3333 here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AURdZitXfR5N9A28a1b2Hp
The committed docs tree had drifted: no pages for getLastGoodBstockPrices or isBstocksDegraded (both added after cc6bd3d regenerated it), nothing for Binance.pricedFresh, and every "Defined in" line number stale after the lint pass reflowed the sources. Regenerating from the main working tree rather than a linked worktree also changes the output: typedoc could not resolve the git remote from a worktree, where .git is a file rather than a directory, so it emitted bare paths. Here it resolves the remote and emits real links -- but by default it pins them to the current commit, which rewrites all 58 pages on every regeneration and leaves the links pointing at an ever-older commit. Setting `gitRevision: "master"` in typedoc.json makes them .../blob/master/src/... instead: stable across regenerations, so a docs refresh only shows real content changes from here on. The trade is that links to code added on a branch resolve once that branch merges. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AURdZitXfR5N9A28a1b2Hp
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds Binance bStocks (tokenized US equities on BNB Smart Chain) as synthetic
market entries in
GET /v2/rates, sourced live from Binance's public SpotAPI. Design spec:
docs/superpowers/specs/2026-08-04-bstocks-integration-design.mdin the ZelTreZ repo (§4.2 covers this repo's contract).
Binanceprovider client (src/services/providers/binance.ts) —singleton, no API key, tokenised-asset list + Spot 24h/7d tickers, with a
last-known-good ticker cache.
getBstockPrices()assembler (src/services/bstocks.ts) — intersectsthe BSC-listed tokenised-asset universe with
TRADING-status Spot symbols,quoted in USDT (verified live — no USDC pairs exist), converts to BTC
via the same ticker batch's
BTCUSDT.zelcoreRatesV2.getAll()and served through the existing/v2/ratesroute.Emitted ids:
bstock-<assetCode lowercase>(e.g.bstock-tslab), alwaystagged
provider: "coingecko"— not"binance". The ZelCore client doesno provider-prefix parsing; it keys its market store on the literal string
${provider}-${id}, and the siblingapirepo advertises each bStock'scoinInfo.coingeckoIDasbstock-<code>. The two literals only meet if theprovider here is exactly
"coingecko"— any other value misses silently,with no error. This id/provider pairing is a cross-repo contract; do not
change it in isolation.
Please read before reviewing the diff
1. The positional-merge fix is the highest-impact change in this PR, and it
affects every provider, not just bStocks. The old
mergeDeepinapiServices.ts'sserviceRefresheroverlaid the freshly-fetchedcryptoarray onto the previous one by array index. That's only correct while
every provider block returns exactly the same row count in the same order.
When a block shrank (a provider outage, a delisted coin), two things broke:
fields from the old entry at that index leaked onto a different coin (e.g.
a CryptoCompare row inheriting CoinGecko's stale
rank/change7d), andentries past the new, shorter length lived on as stale duplicates. Because
the ZelCore client re-keys on
${provider}-${id}with last-write-wins, andthose stale duplicates sat after the fresh ones in the array, wallet users
were served the stale price on any refresh cycle where a provider
block's row count shifted. Reproduced with a concrete input (simulated
CryptoCompare outage: 40 duplicate keys, client resolved to the stale price
instead of the fresh one). Fixed by
replaceCryptoByKey— a key-basedrebuild from the fresh fetch alone (dedup by
${provider}-${id},last-write-wins, no merge with the previous array). Two behaviour changes
worth knowing: repeated
provider+identries collapse to one at the firstoccurrence's position, and an entry the fetch no longer produces disappears
immediately instead of persisting from the previous cycle.
2. The
crypto.length > 300degraded-response guard now counts non-bStockrows only. That floor exists to reject an under-strength provider response
before it overwrites good in-memory data. It was calibrated before bStocks
existed; the ~56 synthetic bStock rows would otherwise pad the count and let
a genuinely degraded CoinGecko/CryptoCompare/LiveCoinWatch response sail
past the guard, truncating
/v2/ratesfor real assets.3. Binance returns a halted symbol present, not absent, during a trading
halt (e.g. around a stock split) — with
lastPrice: "0.00000000". Verifiedlive: 20/20 requested BREAK-status symbols came back present in the ticker
response, 9/20 of those priced at zero. An absent-symbol fallback would have
been dead code for this case, and accepting the zero would both serve a $0
price and poison the last-known-good cache. Prices are therefore accepted
only when finite and strictly positive (
Number.isFinite(px) && px > 0),both for what's served and for what's written to the last-known-good store.
4.
npm run lintis a no-op repo-wide, unrelated to this branch.npx eslint ./ --format jsonlints exactly one file,.eslintrc.jsitself,and zero
.tsfiles —.eslintrc.jsusesparserOptions.parser: babel-eslintwith no TypeScript parser, and ESLint defaults to.jsonly.Every "lint clean" claim in this repo's history is vacuously true.
npx tsc --noEmitis the only real static check and is clean on thisbranch. Flagging as a follow-up; not something this PR fixes.
5. Pre-existing debt this PR does not fix:
mergeDeepis stillpositional for
ratesV2.fiat, v1rates, andmarketsUSDinapiServices.ts— the same stale-tail class of bug described in point 1,just not yet observed to matter for those paths (v1
rates/marketsUSDarevalue-not-array shaped per currency code, so the failure mode is different in
practice;
ratesV2.fiatis a small, rarely-outaged list from a singleprovider). Worth its own ticket rather than folding into this change.
Live verification (real, not simulated)
No automated test proves bStock entries actually reach
/v2/ratesagainstreal Binance, so I booted the service locally and hit the live endpoint.
Environment note: this sandbox has no paid CoinGecko Pro / CryptoCompare /
LiveCoinWatch API keys, and
express-prometheus-middleware'sprom-clientpeer dependency is missing from
package.json(pre-existing, unrelated tothis branch — flagging separately, not fixed here). To get a real listening
server for this verification I temporarily (1) installed
prom-clientlocally and (2) pointed
config.coinGeckoUrlat CoinGecko's free public tierinstead of the paid
pro-apihost. Neither change is in this diff — bothwere reverted before committing; they only existed long enough to prove the
real, unmodified
/v2/ratescode path end-to-end against live Binance +live (free-tier) CoinGecko data.
Result,
curl localhost:3333/v2/rates:bstock-*entries, allprovider: "coingecko", 0 entriestagged
provider: "binance".bstock-tslab:rates.usd 323.87,rates.btc 0.00504911,change24h -0.059,change7d 5.317bstock-nvdab:rates.usd 216.12,rates.btc 0.00336929,change24h 3.511,change7d 9.29bstock-sndkb:rates.usd 1411.78,rates.btc 0.02200954,change24h 6.242,change7d 27.286$327.35 vs
bstock-tslab$323.87 (~1% off); NVDA $215.61 vsbstock-nvdab$216.12 (~0.2% off); SNDK $1,435.69 vs
bstock-sndkb$1,411.78 (~1.7% off— SanDisk was mid-surge on real HBF-standard news at query time, and the
bStock's
change24h/change7dreflect the same move).cryptoarray length: 420 (364 non-bStock + 56 bStock). Non-bStock entries are real, live CoinGecko data (
bitcoin,ethereum,tether,binancecoin,usd-coin, ...) confirming bStocks are additive,not replacing existing coverage.
errors: {"cryptocompare":true, "livecoinwatch":true}reflects the missing paid keys in this sandbox, nota bug in this branch.
Test plan
npx tsc --noEmit— cleannpx jest --runInBand tests/binanceProvider.spec.ts tests/bstocks.spec.ts tests/mergeCrypto.spec.ts— 24/24 passing/v2/ratesverification against real Binance (+ free-tier CoinGecko) — see abovenpm run lint— no-op repo-wide, see point 4 aboveCo-Authored-By: Claude Opus 5 (1M context) noreply@anthropic.com
https://claude.ai/code/session_01UnpQSycV6gfbF9JqRf4x1f