Onvexia — Crypto Fundamentals, Sentiment & Onchain Tracking

The assets with the strongest Galaxy Score right now. Galaxy Score is 0-100 and composite: social volume, engagement, sentiment and market health folded together. A high score is a statement about ATTENTION AND HEALTH, not about valuation — it does not mean an asset is cheap. Call get_asset_scores for the breakdown.

Rank what the market is TALKING ABOUT — themes and narratives, not assets. Ranked by mentions weighted by engagement over the window, so a topic posted about loudly by few accounts does not outrank one discussed widely. Use get_trending_assets for tickers; this is the layer above, where "restaking" and "AI agents" live.

Assets trending now by social activity, with the hype score that separates a real move from a burst of noise.

Current narratives, each an LLM summary of a CLUSTER of posts rather than a single document. Read author_count before quoting one: a high post_count with a low author_count is one person repeating themselves, not a narrative. These are machine summaries, not edited articles.

Recent large on-chain transfers, optionally filtered to one asset. "Whale" is a SIZE threshold, not an identity. A large transfer is very often an exchange moving its own funds between wallets, which is not a market action at all — counterparty labels are included where we hold them, and a null label means WE HAVE NO LABEL, never that the counterparty is unknown or safe.

Cross-wrapper spread for tokenized equities: one real company, every issuer that tokenizes it, and how far apart they trade. SpaceX trades under five wrappers and they do not agree. Spreads run roughly 0.1-0.9 percent between programs referencing the same share. This is a CROSS-WRAPPER comparison, deliberately not a comparison against the underlying stock — that needs a licensed equity feed and no free commercially-usable one exists. A spread is NOT free money. Each issuer carries its own credit, redemption terms and transfer restrictions, and the cheapest wrapper is often cheapest for a reason. Do not present it as an arbitrage.

Browse labelled addresses, filtered by chain, entity (e.g. Binance) or category (exchange | bridge | dex | mev | staking | mining | sanctioned).

List every metric Onvexia knows, with its parity and caveats. parity=exact means we compute it the way Santiment does; approximate means same concept but different coverage or method (read the caveat before relying on the number); unavailable means we do NOT serve it yet. Unavailable metrics are listed on purpose — check here before asserting that Onvexia can answer a question.

Previously compiled natural-language screens: the English somebody wrote and the filter it compiled to, INCLUDING refusals. Useful as worked examples of the screener's filter grammar before you write one.

List the assets that currently have a generated research report available. Returns the index, not the reports — call get_research_report with a symbol for the body. An asset missing from this list has not been written up; that is a statement about our coverage, not about the asset.

Identify a blockchain address — exchange, bridge, DEX, MEV bot, mining pool, or OFAC-sanctioned — with the source and confidence of each label. IMPORTANT: `known: false` means no label is held. It does NOT mean the address is clean or unflagged.

Does a bare ticker actually mean the crypto asset? verdict 'crypto' means mentions are about the asset; 'equity'/'other' means the bare word is dominated by something else (TIA is Spanish 'tia'; GRT collides with 'graph'). A 404 is NOT a clean bill of health — it means nobody has adjudicated that ticker yet.

Run a read-only SELECT against the platform's data. One statement, 15s timeout, 10,000-row cap; truncation is always reported. Call get_sql_schema first for the queryable relations.

Filter the whole asset universe server-side and return the matches. `filter` is comma-separated `field:op:value` terms, e.g. "market_cap:gt:1000000000,funding_rate:lt:0" — assets over $1B whose funding rate is negative. Ops: gt, gte, lt, lte, eq, ne. Call screener_fields() first to see what fields exist and their ranges; guessing a field name gets the whole query rejected.

Every field the screener accepts, with type and description. Call this before building a filter rather than guessing field names.

Search the whole asset universe (~1,900 assets) by symbol or name. Use this to resolve a user's loose reference into a real symbol before calling the other tools.