> For the complete documentation index, see [llms.txt](https://rootai.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://rootai.gitbook.io/docs/mcp/build.md). # Build > rootAI MCP is our flagship product: a free, public, 24/7 market-intelligence backbone for agents, apps, dashboards, bots, research notebooks, Discord bots, Telegram bots, and any system that needs live crypto data plus on-chain context. Plug one URL into your AI assistant or app and skip the data-engineering layer. ``` https://mcp.rootedge.ai/mcp ``` No signup. No API key. No quota application. It is the same production endpoint family that powers the Edge dashboard. When you need premium market intelligence, MCP Pro adds paid, graded Hyperliquid signals across crypto perps and HIP-3 stocks, indices, commodities, FX, pre-IPO assets, and other RWA markets. ``` https://mcp.rootedge.ai/pro ``` Agents can connect through [Base x402](https://www.x402scan.com/server/b5864cea-f1de-4ff0-96dc-0109a7f48a2f), [OKX.AI A2MCP](https://www.okx.ai/agents/2721/), [OpenSea](https://opensea.io/tools/erc8257/base/351) or [Virtuals ACP](https://app.virtuals.io/virtuals/92029). Each path uses the same rootAI signal engine. Discovery, payment, and delivery are the only differences. ### Why build on rootAI MCP * Free and public. No auth, no rate-tier negotiation, no waitlist. * 24/7 production endpoint. Same infrastructure that runs the Edge dashboard. * 98 tools across one protocol. Hyperliquid perps, DexScreener, Polymarket, Binance public, Pyth oracles, Base, Robinhood Chain, TradFi, news, sentiment, and Edge intelligence. * Read-only by default. Discovery is free. Execution-prep tools return unsigned calldata only. * Non-custodial. The MCP never holds keys and never broadcasts transactions on your behalf. * MCP-native. Works with Claude, Cursor, Continue, OpenAI Agents SDK, LangChain MCP adapters, raw JSON-RPC, and anything that speaks Model Context Protocol. * Paid signals on the same endpoint family. MCP Pro adds graded Hyperliquid signals via x402 micropayments, OpenSea, OKX.AI, and Virtuals. ### Use cases * AI trading copilots that read funding, news, signals, and liquidity before suggesting a position. * Research assistants that compare Polymarket odds against on-chain narrative shifts. * Discord or Telegram bots posting daily market context, fear and greed, Hyperliquid movers, and Robinhood Chain pulse. * Custom dashboards stitching Hyperliquid, Pyth, Base launches, Robinhood stock tokens, and DexScreener data. * Backtest harnesses pulling candles, funding, odds, and market context without maintaining separate API clients. * Agent-to-agent systems where rootAI MCP is the shared market truth layer. ### Architecture ``` your client -> https://mcp.rootedge.ai/mcp -> Edge data layer -> live upstreams ``` * Transport: Streamable HTTP with JSON-RPC and SSE responses. * Protocol: MCP `2025-11-25`. * No persistent connection required. Each call is a stateless HTTP request inside an MCP session. * No webhooks. Poll at the cadence your product needs and cache responses on your side. ### 30-second start #### Initialize a session ```bash curl -i https://mcp.rootedge.ai/mcp \ -H "content-type: application/json" \ -H "accept: application/json, text/event-stream" \ -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}' ``` If you get back `serverInfo` and `capabilities`, you are connected. Save the `Mcp-Session-Id` response header and send it with later requests in that session. #### List available tools ```bash curl -s https://mcp.rootedge.ai/mcp \ -H "content-type: application/json" \ -H "accept: application/json, text/event-stream" \ -H "mcp-session-id: " \ -d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' ``` #### Call a tool ```bash curl -s https://mcp.rootedge.ai/mcp \ -H "content-type: application/json" \ -H "accept: application/json, text/event-stream" \ -H "mcp-session-id: " \ -d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"hyperliquid_mids","arguments":{}}}' ``` ### MCP Pro MCP Pro serves rootAI's graded Hyperliquid signals through four access paths: | Path | Best for | | --------------------- | ------------------------------------------------------- | | Base x402 | Custom clients and x402-capable agents | | OpenSea Tool Registry | Users and agents that want the simplest pay-per-call UI | | OKX.AI A2MCP | OKX.AI-compatible agents | | Virtuals ACP | Virtuals marketplace buyers and agents | The signal engine is the same in every path. ### What MCP Pro serves Premium coverage includes Hyperliquid crypto perps and HIP-3 stocks, indices, pre-IPO assets, commodities, FX, and other RWA markets. Signal delivery is deterministic and does not use an LLM. Each signal can include: * asset or coin * asset class * direction or side * signal kind, such as funding divergence, book imbalance, volume surge, OI shift, or liquidation-zone pressure * grade from `A` strongest to `D` weakest * normalized score * confirmation score * tradeability score * context score * concise human-readable interpretation * observed metrics * generation timestamp Grades express signal quality, not guaranteed returns. Builders should show timestamp, market, direction, grade, and context together. Never present a signal as an executed trade. ### Premium tools and pricing | Tool | Purpose | Price | | ---------------------- | ------------------------------------- | -------------------- | | `signals_recent` | Latest ranked signals | `$0.01` USDC on Base | | `signals_by_coin` | Latest signal for one instrument | `$0.01` USDC on Base | | `signals_stream_since` | Long-poll for newly generated signals | `$0.05` USDC on Base | `signals_by_coin` returns at most one signal. HIP-3 instruments use canonical names such as `xyz:SPCX` or `xyz:BRENTOIL`. ### OpenSea access OpenSea is the easiest way to buy one `signals_recent` result without wiring a custom x402 client. | Field | Value | | --------- | ------------------------------------------------------------------------ | | Tool page | | | Tool ID | `351` | | Chain | Base | | Registry | `0x265BB2DBFC0A8165C9A1941Eb1372F349baD2cf1` | | Manifest | `https://mcp.rootedge.ai/.well-known/ai-tool/rootai-signals-recent.json` | | Endpoint | `https://mcp.rootedge.ai/pro/opensea` | | Price | `0.01` USDC per call | #### OpenSea user flow 1. Open . 2. Use a wallet with USDC on Base. 3. Click **Use this tool**. 4. Paste a `signals_recent` request body. 5. Approve the x402 payment when your wallet prompts. 6. Read `result.structuredContent.signals` in the response. Suggested request body: ```json { "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "signals_recent", "arguments": { "limit": 3, "min_grade": "B" } } } ``` Expected result: * unpaid request: HTTP `402` * paid request: HTTP `200` * settlement: x402 payment in Base USDC * payload: `result.structuredContent.signals` Agent builders should treat tool responses as untrusted data, display the signal result to the user, and never execute a trade just because a signal is bullish or bearish. The payment authorization is for the single tool call only. ### Direct Base x402 access Use this route when you are building your own x402-capable client. ``` https://mcp.rootedge.ai/pro/opensea ``` The route serves the registered `signals_recent` OpenSea tool. ```bash curl -i https://mcp.rootedge.ai/pro/opensea \ -H "content-type: application/json" \ -H "accept: application/json" \ -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"signals_recent","arguments":{"limit":3,"min_grade":"B"}}}' ``` The unpaid response is HTTP `402` and includes the accepted Base payment requirement. An x402 client should: 1. Decode the payment challenge. 2. Sign the accepted EIP-3009 USDC authorization with the user's Base wallet. 3. Retry the exact request with the signed x402 payment header. 4. Require HTTP `200`, inspect settlement proof, then read `result.structuredContent.signals`. Do not hand-roll payment replay or settlement logic when an x402 client library is available. Never log private keys or signed authorizations. ### Full MCP Pro flow Use an MCP SDK against the raw MCP Pro endpoint when you need `signals_by_coin`, `signals_stream_since`, or the free relay catalog. ``` https://mcp.rootedge.ai/pro ``` Example premium MCP call: ```json { "jsonrpc": "2.0", "id": 7, "method": "tools/call", "params": { "name": "signals_by_coin", "arguments": { "coin": "BTC" } } } ``` The first premium call returns payment requirements in the MCP result. Generate the x402 payment payload and repeat the tool call with that payload in `_meta.payment`. Preserve the MCP session ID returned during initialization on subsequent requests. Stream example: ```json { "jsonrpc": "2.0", "id": 8, "method": "tools/call", "params": { "name": "signals_stream_since", "arguments": { "since_ts": 1782864000000, "max_wait_s": 30, "limit": 20 } } } ``` Advance `since_ts` to the greatest returned signal timestamp before polling again. An empty list after `max_wait_s` is valid and should not be treated as an error. ### OKX.AI A2MCP | Field | Value | | -------- | --------------------------------------- | | Provider | rootAI on OKX.AI | | Endpoint | `https://mcp.rootedge.ai/pro/okx` | | Network | X Layer, `eip155:196` | | Asset | USD₮0 | | Service | latest ranked signals, `signals_recent` | For OKX.AI agents, open the rootAI provider profile and invoke the listed A2MCP service. OKX.AI handles discovery, authorization, payment, and response delivery. Custom OKX-compatible clients can POST the same `signals_recent` JSON-RPC body used in the Base example to `/pro/okx`. The unpaid response returns HTTP `402` with the X Layer requirement. Sign it with OKX payment tooling, resend the identical body using the returned payment authorization header, and require: * HTTP `200` * a settlement response header * a non-empty `result.structuredContent.signals` The paying wallet needs USD₮0 on X Layer. Do not send Base USDC to the OKX payment route. ### Virtuals ACP | Field | Value | | ---------------- | -------------------------------------------- | | Agent | rootAI on Virtuals | | Provider address | `0x9d23ddd0a527b3b63927956840c8ff35b9db95a6` | | Network | Base, `8453` | | Offering | `rootai_pro_signals` | Virtuals exposes rootAI through its marketplace lifecycle rather than a raw HTTP payment challenge. Supported job requirements: | Field | Meaning | | ------------- | ----------------------------------------------------------------- | | `mode` | `recent` or `by_coin` | | `coin` | required for `by_coin`; use canonical HIP-3 names when applicable | | `limit` | 1 to 10 for `recent`; `by_coin` returns one | | `min_grade` | optional `A`, `B`, `C`, or `D` | | `asset_class` | optional filter such as `crypto` or `stock` | After creation, follow the standard ACP lifecycle: fund the job, wait for submission, inspect the structured deliverable, then complete or reject it. ### Production handling * Validate HTTP status, settlement proof, and response schema before displaying a paid result. * Treat `402` as a payment challenge, not an application failure. * Treat an empty stream response as a successful poll with no new signals. * Make payment retries idempotent. Do not generate a second authorization until the first settlement state is known. * Store timestamps and canonical asset IDs. Do not resolve `xyz:` assets by fuzzy ticker matching. * Prefer grades A and B in compact feeds, but preserve original grade and metrics. * Keep interpretations concise and distinguish signal direction from trade execution. * Apply backoff to `429` and transient `5xx` responses. * Never expose wallet private keys, bearer credentials, payment signatures, or buyer configuration. ## Free MCP tool reference The free MCP endpoint currently exposes 98 public tools. ``` https://mcp.rootedge.ai/mcp ``` Tool families: | Family | Count | Purpose | | ----------------------- | ----: | --------------------------------------------------------------------------------------------------------- | | `hyperliquid_*` | 7 | Hyperliquid public market data and HIP-3 RWA reads | | `polymarket_*` | 37 | Prediction-market discovery, prices, orderbooks, activity, positions, and market metadata | | `dex_*` | 13 | DexScreener token, pair, boost, profile, and narrative discovery | | `binance_public` | 1 | Generic Binance public endpoint wrapper | | `base_*` | 2 | Base unsigned swap and send calldata prep | | `hood_*` | 10 | Robinhood Chain canonical assets, market pulse, stock-token intel, wallet intel, swap prep, and tx status | | `stock_market_cissor_*` | 18 | TradFi quotes, search, history, fundamentals, options, news, and market context | | `edge_*` | 10 | Edge value-add aggregation: launches, fear and greed, perps context, options, news, Pyth | ### `hyperliquid_*` Read-only Hyperliquid market data. Tools: * `hyperliquid_meta` * `hyperliquid_summaries` * `hyperliquid_mids` * `hyperliquid_book` * `hyperliquid_candles` * `hyperliquid_funding` * `hyperliquid_stock_rwa` Use these for perps metadata, mids, orderbook, candles, funding, market summaries, and HIP-3 stock/RWA context. ### `polymarket_*` Prediction-market discovery, pricing, orderbook, flow, and public wallet activity. Discovery: * `polymarket_search` * `polymarket_events` * `polymarket_event` * `polymarket_event_by_slug` * `polymarket_event_tags` * `polymarket_market_tags` * `polymarket_tags` * `polymarket_tag` * `polymarket_series` * `polymarket_series_detail` * `polymarket_sports` * `polymarket_teams` Markets: * `polymarket_markets` * `polymarket_market` * `polymarket_market_by_slug` * `polymarket_market_by_token` * `polymarket_simplified_markets` * `polymarket_clob_markets` Prices and quotes: * `polymarket_price` * `polymarket_prices` * `polymarket_last_trade_price` * `polymarket_last_trade_prices` * `polymarket_midpoint` * `polymarket_midpoints` * `polymarket_spread` * `polymarket_spreads` * `polymarket_tick_size` * `polymarket_prices_history` Orderbook and flow: * `polymarket_orderbook` * `polymarket_orderbooks` * `polymarket_trades` * `polymarket_open_interest` * `polymarket_holders` Public wallet activity: * `polymarket_activity` * `polymarket_positions` * `polymarket_closed_positions` * `polymarket_user_value` Best practice: start with `polymarket_search` using a precise query. Avoid `polymarket_events` as the first move for a specific event. Check market status before quoting odds because resolved markets can return last-known prices that are not live. ### `dex_*` DexScreener token, pair, boost, profile, and narrative discovery. Tools: * `dex_latest_profiles` * `dex_recent_profile_updates` * `dex_latest_community_takeovers` * `dex_latest_ads` * `dex_latest_boosts` * `dex_top_boosts` * `dex_token_orders` * `dex_trending_metas` * `dex_meta` * `dex_pair` * `dex_search` * `dex_token_pairs` * `dex_tokens` Use these for token discovery, pair lookup, liquidity context, boost monitoring, and market-narrative scans. ### `binance_public` One generic public Binance wrapper. Tool: * `binance_public` Example payload: ```json { "path": "/api/v3/ticker/price", "params": { "symbol": "BTCUSDT" } } ``` ### `base_*` Base-chain unsigned calldata prep. Tools: * `base_prepare_swap` * `base_prepare_send` Both return unsigned calls for the caller to sign and submit: ```json { "chain": "base", "calls": [ { "to": "0x...", "value": "0x...", "data": "0x..." } ], "_quote": { "input_amount": "...", "out_values": ["..."], "price_impact": 0.003, "gas_estimate": 203276 } } ``` The MCP never broadcasts. Your app or the user's wallet signs and submits. For one-click UX, use the Edge dashboard's `base_swap` and `base_send` wrappers. ### `hood_*` Robinhood Chain market intelligence and unsigned swap prep. Robinhood Chain details from the live tools: | Field | Value | | ---------------- | -------------------------------------------------------- | | Chain | Robinhood | | Chain ID | `4663` | | RPC health | exposed through `hood_market_pulse` | | Canonical assets | USDG, WETH, 20 stock tokens, 5 ETFs | | Market universe | provider-visible market universe, not a full-chain index | | Explorer | `https://robinhoodchain.blockscout.com` | Tools: | Tool | Purpose | | ----------------------- | ----------------------------------------------------------------------------------------------------------------------------- | | `hood_canonical_assets` | Official Robinhood Chain canonical assets: USDG, WETH, stock tokens, and ETFs | | `hood_stock_intel` | Robinhood stock/ETF token intelligence, ERC-20 metadata, UI multiplier, supply, and oracle pause fields | | `hood_token_intel` | Dynamic token lookup using canonical registry, DexScreener-visible markets, RPC metadata, and capped recent Transfer-log scan | | `hood_wallet_intel` | Wallet summary for native ETH, canonical assets, and optionally provider-visible token holdings | | `hood_trading_context` | Trading context for canonical stocks/ETFs, memecoins, and other Robinhood tokens | | `hood_market_search` | Dynamic provider-visible market search and ranking; use this for top-N market asks | | `hood_market_pulse` | Lightweight chain pulse: latest block, canonical asset counts, and market rankings | | `hood_swap_quote` | LI.FI Robinhood Chain swap quote with validated route and transaction data | | `hood_swap_prepare` | Unsigned Robinhood Chain swap calls, including exact-amount approval if needed | | `hood_tx_status` | Robinhood Chain transaction receipt status | Example: Robinhood Chain pulse ```json { "jsonrpc": "2.0", "id": 10, "method": "tools/call", "params": { "name": "hood_market_pulse", "arguments": {} } } ``` Typical result fields: * `chainId` * `rpc.latestBlock` * `rpc.healthy` * `canonicalAssets.stocks` * `canonicalAssets.etfs` * `canonicalAssets.cashAndGas` * `universeSize` * `markets.topByVolume` * `markets.topByLiquidity` * `markets.fresh` Example: Robinhood market search ```json { "jsonrpc": "2.0", "id": 11, "method": "tools/call", "params": { "name": "hood_market_search", "arguments": { "rank_by": "volume", "limit": 3 } } } ``` Supported `hood_market_search` filters include: * `query` * `rank_by` * `limit` * `category` * `min_volume_24h` * `min_liquidity` * `max_age_hours` Use `hood_market_search` for questions like "top Robinhood Chain tokens by volume" or "fresh Robinhood markets". Use `hood_canonical_assets` for the official stock/ETF/cash registry. Use `hood_swap_prepare` only when the user intends to sign a Robinhood Chain swap. ### `stock_market_cissor_*` TradFi quotes, history, fundamentals, events, options, and market context. Quotes and search: * `stock_market_cissor_search` * `stock_market_cissor_quote` * `stock_market_cissor_multi_quote` History: * `stock_market_cissor_history_daily` * `stock_market_cissor_history_intraday` Market context: * `stock_market_cissor_market_summary` * `stock_market_cissor_sector_data` * `stock_market_cissor_industry_data` Fundamentals: * `stock_market_cissor_fundamentals_summary` * `stock_market_cissor_balance_sheet` * `stock_market_cissor_cash_flow` * `stock_market_cissor_income_statement` Events: * `stock_market_cissor_earnings_dates` * `stock_market_cissor_corporate_actions` * `stock_market_cissor_ticker_news` Options: * `stock_market_cissor_options_chain` * `stock_market_cissor_options_summary` * `stock_market_cissor_options_expirations` ### `edge_*` Edge value-add aggregation. Tools: * `edge_bankr_launches` * `edge_discover_base_token_opportunities` * `edge_fear_greed` * `edge_perps_market` * `edge_options_chain` * `edge_options_spreads` * `edge_news_latest` * `edge_news_by_asset` * `edge_pyth_feeds` * `edge_pyth_price` Use these when you want opinionated market context instead of a raw upstream response: Base launch discovery, fear and greed, perps context, options, latest news, asset-specific news, and Pyth oracle data. ## Built-in prompts The server ships ready-made prompt templates you can invoke directly from any MCP client. Use them when you want a full market read without composing multiple tool calls manually. ## Integration recipes ### Claude Claude web: ``` Customize -> Connectors -> Add custom connector -> https://mcp.rootedge.ai/mcp ``` Claude Desktop config: ```json { "mcpServers": { "rootai": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-fetch"], "env": { "MCP_SERVER_URL": "https://mcp.rootedge.ai/mcp" } } } } ``` ### Cursor ```json { "rootai": { "url": "https://mcp.rootedge.ai/mcp", "type": "sse" } } ``` ### Node / TypeScript ```ts import { Client } from "@modelcontextprotocol/sdk/client/index.js"; import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js"; const transport = new StreamableHTTPClientTransport( new URL("https://mcp.rootedge.ai/mcp") ); const client = new Client({ name: "my-app", version: "1.0.0" }); await client.connect(transport); const tools = await client.listTools(); const result = await client.callTool({ name: "hyperliquid_summaries", arguments: {} }); ``` ### Raw JSON-RPC Use this when your environment does not have an MCP SDK. ```bash curl -s https://mcp.rootedge.ai/mcp \ -H "content-type: application/json" \ -H "accept: application/json, text/event-stream" \ -H "mcp-session-id: " \ -d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"edge_fear_greed","arguments":{}}}' ``` ## Patterns ### Chain tools in one LLM turn Ask for price, funding, odds, news, and signals together when your client supports multi-tool planning. Example: ``` Compare BTC perps funding, recent news, Polymarket election odds, and current fear/greed. Summarize the highest conviction market read. ``` ### Prefer batch and plural tools Use plural/batch tools when available: * `polymarket_prices` * `polymarket_midpoints` * `polymarket_spreads` * `stock_market_cissor_multi_quote` ### Cache on your side The free endpoint is public. Cache stable reads such as metadata, tags, token lists, canonical Robinhood assets, and market IDs. Refresh fast-moving prices, orderbooks, and pulse data more often. ### Handle rate limits gracefully If you receive `429`, back off and retry later. Avoid hammering the endpoint from a single IP. ### Do not double-fetch If an LLM already called `polymarket_market_by_slug`, do not immediately call `polymarket_market` unless you need a different lookup path. ## When you outgrow free MCP Pro adds: * graded Hyperliquid signals * HIP-3 RWA, FX, commodity, stock, index, and pre-IPO coverage * per-call Base USDC through x402 * OpenSea Tool Registry access through Tool 351 * X Layer USD₮0 through OKX.AI * ACP jobs and subscriptions through Virtuals * all 98 free tools mirrored on the Pro endpoint at no cost ## Reference | Item | Value | | -------------------------- | ------------------------------------------------------------ | | Public MCP endpoint | `https://mcp.rootedge.ai/mcp` | | Raw MCP Pro endpoint | `https://mcp.rootedge.ai/pro` | | OpenSea MCP Pro endpoint | `https://mcp.rootedge.ai/pro/opensea` | | OpenSea Tool Registry page | `https://opensea.io/tools/erc8257/base/351` | | Protocol | MCP `2025-11-25` | | Free tools | `98` | | Payment paths | Base x402, OpenSea Tool Registry, OKX.AI A2MCP, Virtuals ACP | | Base chain ID | `8453` | | Robinhood Chain ID | `4663` | | Status | 24/7 production | ## Show off what you build Send a post to `@Root_Edge` and tag your project. We appreciate useful builds, open-source integrations and production case studies. ## Support * Endpoint down or returning errors: send the request and response. * Tool returning an unexpected shape: include the exact `tools/call` payload and `structuredContent`. * Want a new tool: propose it with the workflow it unlocks. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://rootai.gitbook.io/docs/mcp/build.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.