Update to Allium

{
  "name": "product-guide",
  "description": "Guide to Allium's four core products: Explorer, Realtime APIs, Datastreams,\nand Datashares. Explains what each product does, who it's for, and how to\nchoose the right one for a customer's use case.\n\nRead this skill when a customer asks about Allium's products, wants to\nunderstand the difference between them, needs help choosing a product, or\nasks questions like \"which product should I use?\" or \"what does Allium offer?\"\nIt also carries the canonical, dated list of supported blockchains — read it\nfor any \"what chains / how many chains does Allium support?\" question and cite\nthat list instead of answering from memory.",
  "included_files": [
    {
      "relative_path": "references/supported-chains.md",
      "size_in_bytes": 16065
    }
  ],
  "skill_md_contents": "---\nname: product-guide\ndescription: |\n  Guide to Allium's four core products: Explorer, Realtime APIs, Datastreams,\n  and Datashares. Explains what each product does, who it's for, and how to\n  choose the right one for a customer's use case.\n\n  Read this skill when a customer asks about Allium's products, wants to\n  understand the difference between them, needs help choosing a product, or\n  asks questions like \"which product should I use?\" or \"what does Allium offer?\"\n  It also carries the canonical, dated list of supported blockchains — read it\n  for any \"what chains / how many chains does Allium support?\" question and cite\n  that list instead of answering from memory.\n---\n\n# Allium Product Guide\n\nAllium offers four core products for accessing blockchain data. Each serves a\ndifferent access pattern, latency requirement, and integration style. Explorer,\nDatashares, and Datastreams share one historical data platform; the Realtime\nAPIs serve a smaller, separate set of chains.\n\n> **Answering \"what chains does Allium support?\" (grounding rule)**\n>\n> Chain names and counts must come from the canonical list, never from memory —\n> generating them from memory produces unstable totals and invented chains.\n>\n> 1. Fetch the **Allium Supported Chains — Canonical List** reference file (see\n>    \"Reference files\" below) and quote it. For an up-to-the-minute Realtime\n>    figure, call the `realtime_get_supported_chains` tool.\n> 2. Always say **which product** a count refers to — the historical data\n>    platform (Explorer / Datashares / Datastreams) and the Realtime APIs have\n>    different totals.\n> 3. The headline total must equal the number of rows for that product in the\n>    canonical list. Do not round to a vague \"80+/130+/150+\" — cite the exact\n>    count and state the list's snapshot date in the answer, so a reader can see\n>    how current it is (e.g. \"**[N]** chains on the historical data platform, as\n>    of **[snapshot date]**\"). Pull both **[N]** and the date from the fetched\n>    reference file — never from this instruction or from memory.\n> 4. If a chain is not in the canonical list, say it is not currently supported\n>    rather than guessing.\n\n## Product Overview\n\n| Product           | Access Pattern                      | Latency                                | Pricing Model              | Best For                                            |\n| ----------------- | ----------------------------------- | -------------------------------------- | -------------------------- | --------------------------------------------------- |\n| **Explorer**      | SQL queries via web UI or API       | ~1 hour freshness, ~4-5s query time    | Query compute time         | Ad-hoc analysis, dashboards, research               |\n| **Realtime APIs** | REST API (pull)                     | 50-100ms response, 3-5s data freshness | API call volume            | Application backends, wallets, trading UIs          |\n| **Datastreams**   | Push via Kafka/PubSub/SNS/WebSocket | 3-5s data freshness                    | Data destinations & egress | Event-driven architectures, real-time pipelines     |\n| **Datashares**    | Native tables in your warehouse     | 1-3 hours batch, sub-minute streaming  | Number of chains & schemas | Institutional analytics, joining with internal data |\n\n## Explorer\n\n**What it is:** A SQL-based analytics workspace at `app.allium.so/explorer` that\nlets users query, visualize, and share blockchain data across every chain on the\nhistorical data platform (see the canonical chain list for the exact count). Powered\nby Snowflake (OLAP).\n\n**Key capabilities:**\n\n- Full SQL interface with cross-chain queries (e.g., `crosschain.dex.trades`)\n- 1,000+ enriched schemas: raw, decoded, DEX, DeFi, NFTs, stablecoins, wallet 360, metrics, bridges, prices, PnL\n- Interactive chart builder with public sharing and embed links\n- CSV and API data upload — join your own data with on-chain data\n- Parameterized queries for reusable, dynamic SQL\n- AI Assistant for natural-language-to-SQL\n- Explorer API — programmatic query lifecycle (create, run, fetch results, cancel)\n- Curated metrics dashboards for stablecoins, DEXs, bridges, etc.\n\n**Target users:**\n\n- Analytics teams at crypto companies (foundations, protocols, wallets)\n- Data analysts and researchers\n- Audit, accounting, and compliance teams (Big 4 firms)\n- Growth and marketing teams tracking ecosystem metrics\n\n**Common use cases:**\n\n- User behavior analytics and wallet activity patterns\n- Ecosystem metrics and competitive intelligence\n- Sybil detection for airdrop eligibility (used by Jito, Wormhole)\n- Account reconciliation and auditing (used by Big 4 firms)\n- DEX adoption dashboards (Uniswap Foundation)\n- Financial monitoring and tax reporting (TaxBit)\n\n**When to recommend Explorer:**\n\n- Customer wants to explore data interactively with SQL\n- Ad-hoc analysis, research, or one-off investigations\n- Building shareable dashboards and visualizations\n- Uploading proprietary data to join with on-chain data\n- Needs the broadest chain coverage (the full historical data platform; see the\n  canonical chain list) and deepest schema library (1,000+)\n- Hourly data freshness is acceptable\n\n## Realtime APIs\n\n**What it is:** Production-grade REST APIs delivering real-time, enriched\nblockchain data with 50-100ms response times and 3-5 second data freshness.\nRealtime covers fewer chains than the historical data platform — quote the\ncanonical chain list (or the live `realtime_get_supported_chains` tool) for the\nexact set and count.\n\n**Key capabilities:**\n\n- **Prices** — real-time and historical token prices from on-chain DEX trades (not centralized exchanges). OHLC candles, VWAP, z-score outlier filtering. New tokens priced within seconds of first DEX trade (including pump.fun launches)\n- **Tokens** — metadata, type, price, decimals, FDV, volume, trade count, holders, ATH/ATL, liquidity, creation time. Sortable and searchable\n- **Wallets** — current balances (native + ERC-20/SPL), historical balances at any point in time, full transaction history with activity labels (swaps, transfers, DEX trades)\n- **Holdings** — real-time and historical USD portfolio holdings with built-in PnL using average cost basis. Multi-granularity (15s/5m/1h/1d)\n- **Assets** — batch-fetch asset details by chain + address\n- **Hyperliquid** — dedicated endpoints for the Hyperliquid protocol\n- Custom endpoints backed by your own SQL queries\n\n**Performance:**\n\n- 50-100ms response time\n- 1-1.2s raw block freshness, 3-5s enriched data freshness\n- Tested to 90,000 RPS (Phantom/Jupiter airdrop: 477M requests in 4 hours)\n- 99.9% uptime SLA\n\n**Authentication:** API key via `X-API-KEY` header. Generate at Settings > API Keys.\n\n**Target users:**\n\n- Application developers building crypto products\n- Wallet teams (Phantom, MetaMask)\n- DEX/trading platforms (Uniswap, Fomo)\n- Payment providers (MoonPay, Bridge.xyz)\n- Fraud detection systems (Cube3.ai, Blowfish)\n- AI agent builders\n\n**When to recommend Realtime APIs:**\n\n- Customer is building an application that needs to pull data on demand\n- Needs sub-second response times for user-facing features\n- Wallet balances, transaction history, token prices, or portfolio PnL\n- Request-response pattern fits their architecture\n- Needs instant coverage of new tokens (long-tail/pump.fun)\n- Building token screeners, trading interfaces, or portfolio trackers\n\n## Datastreams\n\n**What it is:** Real-time push-based data delivery via enterprise message brokers.\nEnriched, decoded blockchain data from the historical data platform (see the\ncanonical chain list for the exact count) streamed with 3-5 second latency and\nguaranteed delivery.\n\n**Key capabilities:**\n\n- Delivery via **Kafka**, **Google Cloud Pub/Sub**, **Amazon SNS** (coming soon), **WebSockets**, and **webhooks**\n- Guaranteed delivery and message ordering (Kafka/PubSub)\n- Historical replay via retention policies\n- **Stream Transformations** — managed filter-and-route pipelines:\n  - Data source filters (dynamic address/contract lists, updateable without restart)\n  - Declarative JSON filters with `=`, `!=`, `>`, `<`, `in`, `exists`, `AND`/`OR`\n  - Workflows: `source → filter → destination`\n- **Beam** (custom transforms) — JavaScript v8 transforms and redis set filters on any stream, with Kafka/SNS sinks. See `beam-pipelines` skill for details\n- Compression (lz4, zstd, gzip)\n- WebSocket streaming of all Kafka topics for simpler integration\n\n**Data entities:** Raw (blocks, transactions, logs, traces), decoded logs/traces, DEX trades, token transfers, balances — across the historical data platform (see the canonical chain list for the exact count).\n\n**Target users:**\n\n- Teams building event-driven blockchain infrastructure\n- Wallet backends (Phantom)\n- Market intelligence platforms (Messari)\n- Fraud detection engines (Blowfish)\n- Back-office reconciliation systems (Bridge)\n- Trading platforms needing real-time token/trade feeds\n\n**When to recommend Datastreams:**\n\n- Customer needs push-based, event-driven data delivery\n- Building real-time alerts, monitoring, or anomaly detection\n- Wants guaranteed delivery with replay capability\n- Needs to filter high-volume streams to specific contracts, addresses, or events\n- Architecture is built around Kafka, PubSub, or message queues\n- Needs custom transformations on streaming data (→ Beam)\n- Wants data pushed rather than polling an API\n\n## Datashares\n\n**What it is:** Managed delivery of production-ready blockchain data as native\ntables directly into your own data warehouse or data lake. Allium handles bulk\ningestion, incremental updates, schema migrations, reorg handling, and data quality.\n\n**Key capabilities:**\n\n- **Snowflake** — native Data Shares (zero-copy). Primary region: GCP US Central 1. Worldwide delivery with 3-hour freshness\n- **BigQuery** — via Google Analytics Hub. Regions: US Central 1, EU West 2\n- **Databricks** — via Delta Sharing. Sub-minute streaming available. Recommended: AWS us-east-2\n- **Amazon S3** / **Google Cloud Storage** — Iceberg format data dumps\n- Apache Iceberg table format with backward-compatible schema evolution\n- SOC 1 & SOC 2 (Type I & II) certified pipelines\n- Full privacy — Allium cannot see your queries, joins, or results\n- No query metering — you control and pay for your own compute\n- Native BI tool connectors: Tableau, Looker, Power BI, Hex, Sigma, Omni\n\n**Data coverage:** the full historical data platform (see the canonical chain list for the exact count), 1,000+ enriched schemas (raw, decoded, DEX, DeFi, NFTs, stablecoins, wallet 360, metrics, bridges, staking).\n\n**Target users:**\n\n- Institutional analytics teams needing data in their own environment\n- Audit, accounting, and compliance teams at regulated institutions\n- Data science teams building models on blockchain data\n- Companies that must join on-chain data with proprietary internal data\n\n**Notable customers:** Visa, Coinbase, Grayscale, Paradigm, Stripe, Uniswap Foundation, MoonPay, Electric Capital.\n\n**When to recommend Datashares:**\n\n- Customer already has a data warehouse (Snowflake, BigQuery, Databricks)\n- Needs to join blockchain data with internal/proprietary data\n- Privacy and data sovereignty are requirements (regulated industries)\n- Running heavy analytical workloads where controlling compute costs matters\n- Building ML/AI models on blockchain data\n- Needs petabyte-scale historical data for backtesting or research\n- Compliance, audit, or accounting use cases at institutions\n\n## How to Choose the Right Product\n\n### Decision Framework\n\n**Start with the access pattern:**\n\n1. **\"I want to explore and analyze data interactively\"** → **Explorer**\n2. **\"I'm building an app and need to fetch data on demand\"** → **Realtime APIs**\n3. **\"I need data pushed to my systems in real-time\"** → **Datastreams**\n4. **\"I want blockchain data in my own warehouse\"** → **Datashares**\n\n### By Use Case\n\n| Customer Need                                       | Recommended Product  |\n| --------------------------------------------------- | -------------------- |\n| Ad-hoc SQL analysis and dashboards                  | Explorer             |\n| Research and data exploration                       | Explorer             |\n| Shareable charts and visualizations                 | Explorer             |\n| Wallet balances and transaction history for an app  | Realtime APIs        |\n| Token prices for a trading interface                | Realtime APIs        |\n| Portfolio PnL in a consumer product                 | Realtime APIs        |\n| Real-time alerts on smart contract events           | Datastreams          |\n| Streaming DEX trades to an analytics pipeline       | Datastreams          |\n| Custom filtered feeds (specific wallets, contracts) | Datastreams (+ Beam) |\n| Institutional-grade historical analytics            | Datashares           |\n| Joining on-chain + internal data for compliance     | Datashares           |\n| ML model training on blockchain data                | Datashares           |\n| BI dashboards in Tableau/Looker/Power BI            | Datashares           |\n\n### By Latency Requirement\n\n| Requirement               | Product                                                                            |\n| ------------------------- | ---------------------------------------------------------------------------------- |\n| Sub-second response time  | Realtime APIs (50-100ms)                                                           |\n| Low-second streaming      | Datastreams (3-5s)                                                                 |\n| Near-real-time analytics  | Explorer (~1 hour) or Datashares (1-3 hour batch, sub-minute Databricks streaming) |\n| Batch/historical analysis | Explorer or Datashares                                                             |\n\n### By Team Profile\n\n| Team                               | Start With                                                                                   |\n| ---------------------------------- | -------------------------------------------------------------------------------------------- |\n| Data analysts writing SQL          | Explorer                                                                                     |\n| Backend engineers building APIs    | Realtime APIs                                                                                |\n| Infrastructure/platform engineers  | Datastreams                                                                                  |\n| Data engineering / warehouse teams | Datashares                                                                                   |\n| Compliance / audit teams           | Datashares (for privacy + joining internal data) or Explorer (for interactive investigation) |\n\n### Common Multi-Product Patterns\n\nMany customers use multiple products together:\n\n- **Explorer + Datashares**: Explore and prototype queries in Explorer, then productionize with Datashares in their warehouse\n- **Realtime APIs + Datastreams**: APIs for user-facing request-response, Datastreams for backend event processing\n- **Datashares + Explorer**: Datashares for heavy analytics in their warehouse, Explorer for ad-hoc investigation and sharing\n- **Datastreams + Datashares**: Datastreams for real-time event processing, Datashares for historical backfill and batch analytics\n\n### Pricing Comparison\n\n| Product       | Model                 | Implication                                        |\n| ------------- | --------------------- | -------------------------------------------------- |\n| Explorer      | Query compute time    | Cost scales with query complexity and frequency    |\n| Realtime APIs | API call volume       | Cost scales with request volume                    |\n| Datastreams   | Destinations & egress | Cost scales with number of streams and data volume |\n| Datashares    | Chains & schemas      | Predictable cost based on data coverage selected   |\n\nAll products require contacting Allium for specific pricing. Customers can sign up for a free trial at `app.allium.so/join` for API access, or book a demo for enterprise needs.\n"
}