Ship: anti-cache RPC probe, keyed bench 069, indexing-freshness 070, Monad+MegaETH 071-072, explorer coverage 068

.github/workflows/staging-deploy.yml

@@ -46,7 +46,11 @@ jobs:
       - name: Deploy Preview
         id: deploy
         run: |
-          url=$(vercel deploy --prebuilt --token=${{ secrets.VERCEL_TOKEN }})
+          # --archive=tgz uploads ONE tarball instead of thousands of
+          # files: the team's api-upload-paid quota (40k reqs / rolling
+          # 24h) was exhausted twice on busy merge days (2026-07-06/07),
+          # blocking every deploy for hours.
+          url=$(vercel deploy --prebuilt --archive=tgz --token=${{ secrets.VERCEL_TOKEN }})
           echo "url=$url" >> "$GITHUB_OUTPUT"
           echo "Preview ready at $url"
 

.gitignore

@@ -32,6 +32,7 @@ yarn-error.log*
 
 # env files (can opt-in for committing if needed)
 .env*
+!.env.example
 
 # vercel
 .vercel
@@ -58,6 +59,8 @@ next-env.d.ts
 
 # compiled harness binaries (go build artifacts)
 harnesses/*/script
+# hl-archive's script/ is a Go source package, not a binary
+!harnesses/hl-archive/script/
 harnesses/*/monitor
 harnesses/*/cmd/script/script
 harnesses/*/cmd/monitor/monitor

alternatives/README.md

@@ -41,6 +41,7 @@ status: live                            # optional, default live
 | `bitquery` | Bitquery | network-coverage |
 | `birdeye` | Birdeye | metadata-coverage (Solana only) |
 | `relay` | Relay | bridge-quote-latency |
+| `zerion` | Zerion | portfolio-chain-coverage |
 
 ## License
 

alternatives/alchemy.yml

@@ -7,5 +7,5 @@ benchmark: aggregator-head-lag
 intro: |
   Alchemy is an EVM-first node and API provider whose surface is shaped by its enhanced endpoints (`alchemy_getAssetTransfers`, NFT API, webhooks, getTokenBalances, the simulation and debug suite, the account-abstraction bundler) rather than by a raw RPC tuned for edge latency. Pricing is metered in compute units, which charges more for an enhanced call than for a plain `eth_call`, so the ceiling shows up first on teams streaming high-cardinality events into a backend rather than on dapp front-ends issuing single reads. Solana coverage exists but sits on a stack built around Ethereum and its L2s. Teams that need a live view of trades on Base, BNB Chain or Solana usually leave Alchemy at the point where the enhanced API list stops mapping to what the product actually reads from chain.
 
-seo_title: Alchemy alternatives. live head-lag benchmark across data providers
+seo_title: "Alchemy alternatives: live benchmark"
 seo_description: Compare Alchemy alternatives on real-time data freshness. Live head-lag against a canonical archive node, refreshed every minute and published openly.

alternatives/birdeye.yml

@@ -8,5 +8,5 @@ chain: solana
 intro: |
   Birdeye is a Solana token analytics platform with a price + metadata API. If you need an alternative, one practical axis is metadata completeness. for a freshly-launched Solana token, how often does the API return a populated logo, description, twitter and website? Below is the live percentage for each provider on Solana, refreshed every minute against fresh-launch tokens detected from Mobula Pulse.
 
-seo_title: Birdeye alternatives. live Solana metadata coverage benchmark
-seo_description: Compare Birdeye alternatives on token metadata coverage for Solana. Logo, description, twitter and website availability for freshly-launched tokens, measured live.
+seo_title: "Birdeye alternatives: live benchmark"
+seo_description: "Birdeye alternatives on Solana token metadata coverage: logo, description, socials in seconds. Codex vs Mobula live."

alternatives/bitquery.yml

@@ -7,5 +7,5 @@ benchmark: network-coverage
 intro: |
   Bitquery is a GraphQL surface on top of indexed onchain data, with the schema as the actual product: queries declare exactly the fields needed across transfers, trades, DEX events and account state, and the indexer fans those out across chains. Pricing is point-based, with each field, filter and join consuming from a monthly point budget rather than from a flat request quota, which means a single dashboard with many panels can hit the ceiling well before request volume looks high. Chain coverage is broad but uneven in depth: some networks expose full DEX trade indexing, others only basic transfers. The team usually compared next is the one whose chain list either covers a long-tail network Bitquery does not, or covers the same network with a flatter pricing model that does not penalise wide GraphQL fragments on a per-field basis.
 
-seo_title: Bitquery alternatives. live network coverage benchmark
+seo_title: "Bitquery alternatives: live benchmark"
 seo_description: Compare Bitquery alternatives on the number of blockchains each major onchain data provider officially supports. Live data, refreshed every six hours.

alternatives/chainlink.yml

@@ -7,7 +7,7 @@ benchmark: oracle-deviation
 intro: |
   Chainlink publishes on-chain price feeds via AggregatorV3 contracts deployed on Ethereum mainnet and most major EVM chains, with node operators posting new rounds on either a deviation trigger (typically 0.25 to 0.5 percent for blue chips) or a heartbeat interval. Every on-chain product that settles in fiat (perp DEXes, lending markets, derivatives, stablecoin issuers) picks one oracle and inherits its drift, which is why the practical question is how aligned the major feeds actually are at the moment a settlement happens. This benchmark polls four oracles every 30 seconds for ten USD-quoted blue chips, computes the pairwise deviation `|a-b|/avg * 100` across every source pair, and publishes the per-pair maximum as a Prometheus gauge in basis points. Sources include Chainlink AggregatorV3 (via `eth_call` against the canonical mainnet contracts), Pyth Network's Hermes REST feed, Binance and Coinbase REST tickers. Pairs are BTC, ETH, SOL, BNB, AVAX, LINK and POL on all four sources, plus XRP, ADA and DOGE on three sources because the Chainlink mainnet feeds for those USD pairs are deprecated. A stale-price guard skips samples older than 60 seconds so a dead poller cannot artificially flatten the deviation, and the companion gauge `ocb_oracle_last_round_age_seconds{source="chainlink"}` surfaces Chainlink's on-chain update gap separately so a wide deviation can be attributed to the right cause. The headline leaderboard ranks the ten pairs by p99 of the max deviation over 24 hours, lower meaning tighter cross-oracle alignment.
 
-seo_title: Chainlink alternatives. Live oracle deviation benchmark in basis points
-seo_description: Compare Chainlink against Pyth, Binance and Coinbase on live oracle deviation. Max pairwise gap on 10 USD pairs, polled every 30 seconds, p99 over 24 hours.
+seo_title: "Chainlink alternatives: live benchmark"
+seo_description: "Chainlink vs Pyth, Binance and Coinbase on live oracle deviation. Max pairwise gap on BTC, ETH, SOL."
 
 status: live

alternatives/coingecko.yml

@@ -7,5 +7,5 @@ benchmark: network-coverage
 intro: |
   CoinGecko is a coin-centric price and market data REST API built around a curated asset list (the `/coins` universe) rather than around on-chain events. Tokens are added through a listing process, prices land on the API after that process completes, and the free tier is famously throttled (around 30 requests per minute on the demo key) which pushes any real product onto a paid plan within days of integration. The on-chain DEX side ships as a separate `/onchain` namespace inherited from GeckoTerminal, with its own quotas. Teams usually leave CoinGecko at one of two points: when the launch cadence of the tokens they care about outpaces the listing pipeline, or when the per-minute cap on the free tier forces a migration before the product is monetised enough to justify the enterprise tier.
 
-seo_title: CoinGecko alternatives. live network coverage benchmark
+seo_title: "CoinGecko alternatives: live benchmark"
 seo_description: Compare CoinGecko alternatives on the number of blockchains each major onchain data API officially supports. Live data, refreshed every six hours.

alternatives/drpc.yml

@@ -7,7 +7,7 @@ benchmark: rpc-capabilities
 intro: |
   dRPC is a decentralized RPC mesh that routes each JSON-RPC request across third-party node providers with consensus checks, exposing a no-key public tier on every major EVM network. Picking a default endpoint for a dapp comes down to a per-chain question: which free public RPC is actually fast and reliable on the chain my product runs on, right now, not according to a marketing page. This benchmark answers it directly. The harness sends an identical `eth_blockNumber` POST every 15 seconds against 15 audited no-key public RPC providers across 10 EVM chains (Ethereum, Arbitrum, Base, Optimism, Avalanche, BNB, Polygon, Linea, Scroll, Mantle), from three Railway regions (us-east, eu-west, sgp). p50, p90 and p99 are computed via Prometheus `quantile_over_time` over the last 24 hours, and a five-class result label (`ok`, `http_err`, `jsonrpc_err`, `stale`, `timeout`) lets the success-rate column catch endpoints that return HTTP 200 with a JSON-RPC error body (the Cloudflare-eth trap). An archive-depth probe runs every 5 minutes against five depths so a fast endpoint that silently serves pruned state is not ranked as a viable archive source. The cohort competing with dRPC includes PublicNode, 1RPC, MeowRPC, Tenderly Gateway, Nodies, Lava, Merkle, Flashbots Protect and Cloudflare, plus the chain-official foundation endpoints on Base, BNB, Arbitrum, Optimism and Avalanche. The chain tab filters every query so the leader is reported per network rather than as a cross-chain aggregate that rewards whichever single-chain endpoint happens to be fastest on its one chain.
 
-seo_title: dRPC alternatives. Live free public RPC benchmark across EVM chains
-seo_description: Compare dRPC alternatives on eth_blockNumber p50 latency. PublicNode, 1RPC, Tenderly, Nodies, Lava and chain-official endpoints measured every 15 seconds from 3 regions.
+seo_title: "dRPC alternatives: live benchmark"
+seo_description: "dRPC alternatives on eth_blockNumber p50 latency. PublicNode, 1RPC, Tenderly, Nodies, Lava ranked live."
 
 status: live

alternatives/dune.yml

@@ -7,5 +7,5 @@ benchmark: wallet-labels-coverage
 intro: |
   Dune lets analysts query indexed blockchain data with SQL, often via labelled wallet datasets contributed by the community. When sizing alternatives, one practical axis is wallet labelling depth. how many addresses does each provider have a label for, and how confident are those labels? Below is the live coverage from each major labelling provider, refreshed continuously against a shared sample of recently active wallets.
 
-seo_title: Dune alternatives. live wallet-labels coverage benchmark
-seo_description: Compare Dune alternatives on wallet labelling depth. Coverage of recently active addresses across major providers, measured continuously and published openly.
+seo_title: "Dune alternatives: live benchmark"
+seo_description: "Dune alternatives on wallet labelling depth. Recently active address coverage across major EVM chains."

alternatives/dydx.yml

@@ -7,7 +7,7 @@ benchmark: perp-fees
 intro: |
   dYdX v4 is a Cosmos-based decentralized perpetuals exchange that runs an on-chain orderbook on its own appchain, with a published tier-0 taker fee of 5 basis points and one of the tightest non-zero-fee books outside Hyperliquid. Published fee schedules are only half the story for a trader picking a venue, because the rack rate ignores the spread crossed at market and the price impact eaten at notional. This benchmark measures the live all-in cost of opening a $1000 ETH long 10x position across major perpetual venues, every five minutes, by reading taker fees from each venue's own API (no hardcoded schedules) and walking each orderbook for $1000 of buy-side notional to compute the spread plus impact component. No transactions are sent. dYdX is queried via its indexer at `/orderbooks/perpetualMarket/ETH-USD` and `/perpetualMarkets`, plus the Cosmos REST endpoint `/dydxprotocol/v4/feetiers/perpetual_fee_params` for the tier-0 default fee. The same harness reads Hyperliquid via `userFees` and `l2Book`, GMX v2 through the synthetics-arbitrum subgraph for `positionFeeFactor`, Lighter via `/orderBookDetails` and `/orderBookOrders`, and gains.trade on Base by reading the fee directly on-chain. All-in cost equals taker fee plus spread, both surfaced as separate metrics. Funding rate is recorded separately (`perp_fees_funding_rate_per_hour_bps`) because it accrues per hour held rather than at open, so bundling it into a single one-number leaderboard would conflate two different fee mechanics. The dimension tab at the top of the page swaps the underlying asset between ETH, BTC and SOL.
 
-seo_title: dYdX alternatives. Live all-in perp fee benchmark on a $1000 ETH 10x long
+seo_title: "dYdX alternatives: live benchmark"
 seo_description: Compare dYdX v4 against Hyperliquid, Lighter, GMX v2 and gains.trade on live taker fee plus spread crossed at $1000 notional, refreshed every five minutes.
 
 status: live

alternatives/etherscan.yml

@@ -7,7 +7,7 @@ benchmark: gas-estimation
 intro: |
   Etherscan is the most-visited block explorer on the web and publishes a gas tracker exposing `SafeGasPrice`, `ProposeGasPrice` and `FastGasPrice` via the v2 API on Ethereum and Polygon (free tier). The tiering was originally designed for the pre-EIP-1559 single-price world and is now mapped onto a priority-fee comparison alongside dedicated oracles. The question every wallet, swap router and bridge UI faces before sending a transaction is which gas oracle actually matches what the next block will charge on the chain their product runs on, rather than what a marketing page claims. This benchmark polls each oracle at its tier-tolerant cadence (Etherscan every 15 seconds with a global rate-gate enforcing at least 6 seconds between any two Etherscan requests across chains, because the no-key limit is 1 req per 5 s per IP shared), buffers the prediction with the predicted block height, and when that block is mined pulls the full block via `eth_getBlockByNumber(.., true)` on the chain's PublicNode RPC. Realized percentiles are computed from actual `maxPriorityFeePerGas` values, and the absolute error per (oracle, tier, chain) lands on both a gauge and a histogram. The ranking metric is the p99 gap over 24 hours, the worst 1 percent of blocks, because typical-minute gaps sit within fractions of a micro-gwei (economically indistinguishable noise) while gas spikes are where a wrong prediction either overpays or misses the block. A covered-rate column shows the share of time each prediction sat at or above the realized p50, the inclusion-side risk an absolute gap cannot show. The cohort competing with Etherscan includes PublicNode `eth_feeHistory` (thin wrapper over EIP-1559 reward percentiles) and Owlracle (multi-oracle aggregator).
 
-seo_title: Etherscan alternatives. Live gas oracle accuracy benchmark
+seo_title: "Etherscan alternatives: live benchmark"
 seo_description: Compare Etherscan against PublicNode feeHistory and Owlracle on gwei gap between predicted and realized priority fee, ranked on p99 over 24h.
 
 status: live

alternatives/gmx.yml

@@ -7,7 +7,7 @@ benchmark: perp-fees
 intro: |
   GMX v2 is an oracle-priced perpetuals venue on Arbitrum that settles against synthetic pools rather than a central orderbook, with a position fee split into a positive-impact branch (4 basis points, reduces venue skew) and a negative-impact branch (6 basis points, adds to skew). Which branch fires depends on open interest at the moment of trade, not on user input, so the worst-case open is the right conservative figure to compare against an orderbook venue's all-in cost. This benchmark measures the live all-in cost of opening a $1000 ETH long 10x position across major perpetual venues, every five minutes, by reading the variable position fee from the synthetics-arbitrum subgraph (`positionFeeFactorForNegativeImpact`) and computing the all-in figure in basis points. No transactions are sent. The same harness queries Hyperliquid (`userFees` plus `l2Book`), dYdX v4 (Cosmos REST plus indexer), Lighter (`/orderBookDetails` plus `/orderBookOrders`), and gains.trade on Base where the fee is read directly on-chain via `eth_call pairs(N)` and `fees(feeIndex).openFeeP`. Because GMX is oracle-priced, the all-in number has no orderbook spread component; the rack rate effectively becomes the figure, which differs structurally from Hyperliquid and Lighter where the half-spread plus impact at $1000 notional gets layered on top of the taker fee. Funding rate is published separately (`perp_fees_funding_rate_per_hour_bps`) because it accrues per hour held, not at open. The asset tab at the top of the page swaps between ETH, BTC and SOL.
 
-seo_title: GMX alternatives. Live all-in perp fee benchmark on a $1000 ETH 10x long
+seo_title: "GMX alternatives: live benchmark"
 seo_description: Compare GMX v2 against Hyperliquid, dYdX, Lighter and gains.trade on live taker fee plus spread crossed at $1000 notional, refreshed every five minutes.
 
 status: live

alternatives/hyperliquid.yml

@@ -7,7 +7,7 @@ benchmark: perp-fees
 intro: |
   Hyperliquid is an L1 perpetuals exchange running on HyperBFT consensus, with an on-chain orderbook and a published base taker fee of 4.5 basis points. Its book is among the deepest of any decentralized venue, which means the spread crossed at market and the impact eaten at $1000 notional usually stay compressed and the all-in cost tracks the rack rate closely. This benchmark measures the live all-in cost of opening a $1000 ETH long 10x position across major perpetual venues, every five minutes, by reading taker fees from each venue's own API and walking each orderbook for $1000 of buy-side notional. No transactions are sent. Hyperliquid is queried via `POST /info {type: l2Book}` for asks, `{type: userFees, user: 0x000...000}` for taker fee and `{type: metaAndAssetCtxs}` for funding. The same harness queries Lighter (`/orderBookDetails` plus `/orderBookOrders`), dYdX v4 (indexer plus Cosmos REST), GMX v2 (synthetics-arbitrum subgraph for `positionFeeFactor`), and gains.trade on Base (fee read directly on-chain via `eth_call`). All-in cost equals taker fee plus half-spread plus impact, all in basis points, with both components emitted as separate Prometheus gauges for transparency. Funding rate is published separately (`perp_fees_funding_rate_per_hour_bps`) because it accrues per hour held rather than at open, so a trader sizing a multi-hour position can layer it on top of the open cost. The dimension tab at the top of the page swaps the underlying asset between ETH, BTC and SOL, with the leaderboard re-sorting on each switch.
 
-seo_title: Hyperliquid alternatives. Live all-in perp fee on a $1000 ETH 10x long
+seo_title: "Hyperliquid alternatives: live benchmark"
 seo_description: Compare Hyperliquid against Lighter, dYdX, GMX v2 and gains.trade on live taker fee plus spread crossed at $1000 notional, refreshed every five minutes.
 
 status: live