This service is public MARKET DATA ONLY. It contains no wallet address, positions, credentials, order signing or trading functions. It has no scheduled jobs. It was verified with actual Hyperliquid responses on 2026-09-05, not only mocked tests.

• Val: kamibaba/hyperliquid-market-bridge
• HTTP file: index.tsx; branch: main
• Bridge version: 1.1.1
• Fixed upstream: https://hyperliquid-readonly-proxy.netlify.app
• Netlify project: hyperliquid-readonly-proxy, site ID 378ad326-88f1-4aef-aac5-1a3734fe8627
• No custom domain is needed. The unrelated sound-notes.net project must not be edited.

Use the connected Val_Town.fetch_val_endpoint tool. Always pass the actual val identity and HTTP file; do not construct or guess val endpoint URLs.

1. Fetch all venues: { "val": "kamibaba/hyperliquid-market-bridge", "path": "index.tsx", "search": "mode=market&limit=0" }.
2. Validate HTTP status, data.scope.coverageComplete, registryVerified, active/valid counts, venue errors and data.fetchedAt. The universe is dynamic, not a fixed 315 or 370.
3. Read every active row without refetching upstream: mode=read&data=market&offset=0&limit=100; then 100, 200, 300 until hasMore=false. If tool output truncates, reduce limit. Check fetchedAt is unchanged across pages.
4. Inspect shortlisted symbols including interval deltas: mode=read&data=market&symbols=HYPE,ZEC,NEAR,DASH,LIT. Keep selection small to avoid output truncation. This is explicitly a subset of the already-fetched full market.
5. Fetch closed candles and book for at most two symbols per call: mode=technical&coins=HYPE,BTC or mode=technical&coins=xyz:MU,xyz:AVGO. Default response is compact; raw is available using view=raw.
6. If a response is truncated or interrupted, mode=read&data=technical reads the saved result without an upstream call. Verify data.symbols matches the intended pair before using it. A 429 may coexist with a recently completed request: inspect the stored snapshot rather than blindly retrying.
7. Refresh market or selected technical data near the final trading decision. mode=read is historical snapshot access, never a promise of current market data.

The connector's built-in Val_Town.web_fetch also successfully fetched the Netlify /health endpoint, but health alone does not verify market coverage.

This val is on the Free tier. The upstream Netlify site remains on its existing Free plan; shared account usage still applies. No paid plan, auto recharge, inference model, cron or trading agent is configured. Additional tool calls may use ChatGPT limits separately.

• Market upstream cooldown: 65 seconds; technical upstream cooldown: 31 seconds.
• Reuse same-query snapshots younger than 60 seconds.
• An application-level budget limits upstream attempts to 200/day (UTC). This is a best-effort application counter, NOT an atomic billing cap or a promise of unlimited free usage. Netlify also enforces its own request limits and Free plan cap.
• Read pages use the stored snapshot, not more Hyperliquid/Netlify collection calls.
• Stored snapshots are overwritten; there is no unbounded full-snapshot archive here.

• All-perp scope includes Core, XYZ and every HIP-3 venue returned by the official registry. kind=all also fetches spot metadata/contexts, but this bridge's default rows are active perps; do not claim deep analysis of all spot pairs.
• fetchedAt is collection time. Asset contexts have no independent exchange event timestamp. The order book has its own source timestamp.
• 24-hour returns are rolling returns, NOT the return since midnight JST.
• Impact spread is NOT best bid/ask spread; use the book for execution analysis.
• OI base and OI notional are distinct. OI increase does not identify which directional side initiated trading.
• Missing interval history remains unknown; no inferred liquidation maps or spot CVD.
• Curated tags are incomplete and may overlap. ZEN is a Base token; grouping it with historical privacy names is a thematic comparison, not a same-chain assertion. NEAR also has privacy-related products despite its basic layer1 label.
• All technical indicators use closed contiguous bars. ATR is neither a guaranteed range nor an automatic minimum stop distance.
• Reject partial, stale or wrong-symbol detailed responses. liveForExecution only checks snapshot age, not profitability or fill guarantees.
• Use market snapshots for setup screening, then check specific execution prices and expiry conditions. Do not enlarge risk merely to target a daily dollar profit.

2026-09-05T05:15:51.880Z and 2026-09-05T05:21:23.682Z: all 11 discovered venues returned HTTP 200; 512 listed perp records, 315 active/valid, 326 spot pairs. All 315 active rows were read from the first snapshot in four pages. The second snapshot refreshed the full universe and ranking outputs.

Technical requests succeeded for HYPE, BTC, NEAR, ZEC, DASH, LIT, CASHCAT, ZRO, SKR and ZORA. HYPE/ZEC were refreshed at 2026-09-05T05:23:28.897Z. These are historical verification times, never current trading prices in a future chat.

Market collection does not use microfn. microfn remains a separate account-checking route only when the user reports position/order changes. Do not use this public val to store the user's account data.