Pyth Pro provides real-time and historical financial market data through natural-language queries, including cryptocurrency, equities, FX, and commodities prices. Retrieve price feeds, inspect asset metadata, and compare market prices to support trading, financial analysis, and market research.
Claude is Anthropic's AI assistant, available as a desktop app and in Cowork. It supports custom connectors, so it can call MCP servers directly from a chat.
This guide connects Pyth MCP to Claude Desktop through MewCP, so Claude Desktop can use Pyth MCP's 6 tools, like get_symbols, convert_date_to_timestamp and get_candlestick_data, without your Pyth MCP credentials ever reaching the app.
Your Pyth MCP API key is stored encrypted and injected only at call time. Your agent never holds it, so it can never leak it, log it or paste it into a prompt.
More apps
Access cryptocurrency market data, portfolios and trading operations on Binance. Monitor assets, retrieve market information and automate crypto trading workflows.
Coinfuty provides cryptocurrency market intelligence for tracking digital asset prices, analyzing market trends, monitoring trading activity, and researching crypto opportunities to support data-driven investment decisions.
Access prediction market data, market positions and trading activity on Polymarket. Monitor markets, retrieve information and interact with forecasting-based financial products.
Trade, check margins, and review orders, portfolio, and fundamentals on Upstox — without leaving your agent. Retrieve user profile and fund/margin balances, estimate pre-trade margin across multiple instruments, look up order books and historical trades, review portfolio positions and holdings, pull company fundamentals by ISIN, and report trade-wise profit and loss by segment and financial year.
Other clients
Pyth MCP works the same way in every client below. Pick yours for its own guide.
Free to start.
Scope access to the exact Pyth MCP tools and accounts you allow, work and personal side by side. Anything outside that scope is refused at call time, even when the model asks for it.
Each call is logged: what ran, on which account, and whether it worked. Failures come back structured, so the agent knows whether to retry instead of guessing.
Every app sits behind one secure MCP URL, and there is no server for you to deploy or keep running. Add an integration in the dashboard and it is live on the next call, with no config to change and no new credentials to wire up. Tools load only when a task needs them, so your agent stays fast and focused however many apps you connect.
In Claude, open Settings → Connectors, then choose Add custom connector.
Name it "mewcp" and paste the URL.
A browser window opens the first time the client connects. Sign in and approve access. No API key needed.
Start a new chat and ask: "List the tools you can use from MewCP." You should see your connected apps.
List available Pyth Pro price feeds. Use this FIRST to discover what feeds exist before calling get_latest_price, get_historical_price, get_price_range or get_candlestick_data (those also accept bare pairs like BTC/USD). Filter by asset_type (e.g. crypto, equity, fx, metal, commodity, interest-rate, funding-rate, kalshi) or search by name/symbol. Narrow further with instrument_type (spot, future, ...) or symbol_chain_id (all contracts of one futures chain, e.g. VX). Retired (inactive) feeds are hidden unless include_inactive is true. Returns compact feed metadata: symbol, name, description, pyth_lazer_id, asset_type, instrument_type, state, exponent, min_channel (fastest channel the feed supports), quote_currency, and where set groups (entitlement groups that gate the feed), expiration_time and symbol_chain_id (futures). Pass verbose: true for every catalog field, e.g. market_sessions (trading-hours schedules) and corporate_actions (e.g. stock splits). With an access token, the list also includes feeds visible only to Pro keys, and each feed has `entitled`: true means this key can query the feed right now. When `entitled` is false, `not_entitled_reason` says why: `not_live (...)` means the feed is in beta, not live yet, or retired, so no key can query it (NOT a plan limitation), `requires one of entitlement groups: ...` means the user's plan lacks that entitlement.
Convert a date string to Unix timestamp for use with get_historical_price, get_price_range and get_candlestick_data. Use this tool when you need to compute timestamps — do not calculate Unix timestamps manually. Accepts ISO 8601 dates (e.g. '2026-01-01', '2026-01-01T12:00:00Z'). Returns seconds, microseconds, ISO 8601, and whether the date is within the available data range (April 2025 onward).
Fetch OHLC candlestick bars for one symbol, for charting, technical analysis or backtesting. OHLC values are already human-readable (no exponent). Requires the user's own Pyth Pro access token. If the user configured it in their MCP client, omit `access_token`; otherwise pass it as `access_token`, and if the tool reports it missing, ask the user for it. Symbols can be full names from get_symbols (e.g. 'Crypto.BTC/USD', 'Equity.US.AAPL/USD') or bare pairs like 'BTC/USD'. A bare pair resolves to the live spot feed when there is one, otherwise to the only remaining live match; `resolved_symbols` in the response shows what each input resolved to, and an ambiguous input returns an error listing the candidates. `from` and `to` are Unix seconds only. Resolutions: 1/5/15/30/60 minutes, 120/240/360/720 (multi-hour), D (daily), W (weekly), M (monthly). At most 500 candles are returned; `truncated: true` means narrow the range or use a larger resolution. Historical data is available from April 2025 onward; do not request times before that. Price tools: get_latest_price for now, get_historical_price for one past instant, get_price_range for every update within up to 60 seconds, get_candlestick_data for OHLC bars over longer periods. Timestamp reference (Unix seconds): 2025-04-01 (earliest available) = 1743465600 2026-01-01 = 1767225600 2026-06-01 = 1780272000 Always double-check your timestamp math (year-boundary errors are common), or use convert_date_to_timestamp.
Get the price of one or more feeds at one past instant. Requires the user's own Pyth Pro access token. If the user configured it in their MCP client, omit `access_token`; otherwise pass it as `access_token`, and if the tool reports it missing, ask the user for it. Symbols can be full names from get_symbols (e.g. 'Crypto.BTC/USD', 'Equity.US.AAPL/USD') or bare pairs like 'BTC/USD'. A bare pair resolves to the live spot feed when there is one, otherwise to the only remaining live match; `resolved_symbols` in the response shows what each input resolved to, and an ambiguous input returns an error listing the candidates. If both price_feed_ids and symbols are provided, only price_feed_ids are used. The timestamp accepts Unix seconds, milliseconds or microseconds (auto-detected) and is rounded down to the channel rate (e.g. a multiple of 200 ms for fixed_rate@200ms). Historical data is available from April 2025 onward; do not request times before that. Requested feeds that return no price are listed in `missing_feed_ids` (e.g. beta or coming_soon feeds, or a channel faster than the feed's min_channel). Prices are integers with an exponent field: human-readable price = price * 10^exponent. Pre-computed display_* fields (display_price, display_bid, display_ask, display_confidence, display_ema_price, display_ema_confidence, display_funding_rate) apply the exponent for you. Price tools: get_latest_price for now, get_historical_price for one past instant, get_price_range for every update within up to 60 seconds, get_candlestick_data for OHLC bars over longer periods. Timestamp reference (Unix seconds): 2025-04-01 (earliest available) = 1743465600 2026-01-01 = 1767225600 2026-06-01 = 1780272000 Always double-check your timestamp math (year-boundary errors are common), or use convert_date_to_timestamp.
Get the most recent real-time price for one or more feeds. Requires the user's own Pyth Pro access token. If the user configured it in their MCP client, omit `access_token`; otherwise pass it as `access_token`, and if the tool reports it missing, ask the user for it. Symbols can be full names from get_symbols (e.g. 'Crypto.BTC/USD', 'Equity.US.AAPL/USD') or bare pairs like 'BTC/USD'. A bare pair resolves to the live spot feed when there is one, otherwise to the only remaining live match; `resolved_symbols` in the response shows what each input resolved to, and an ambiguous input returns an error listing the candidates. If both price_feed_ids and symbols are provided, only price_feed_ids are used. Requested feeds that return no price are listed in `missing_feed_ids` (e.g. beta or coming_soon feeds, or a channel faster than the feed's min_channel). Prices are integers with an exponent field: human-readable price = price * 10^exponent. Pre-computed display_* fields (display_price, display_bid, display_ask, display_confidence, display_ema_price, display_ema_confidence, display_funding_rate) apply the exponent for you. Price tools: get_latest_price for now, get_historical_price for one past instant, get_price_range for every update within up to 60 seconds, get_candlestick_data for OHLC bars over longer periods.
Get every price update for one or more feeds within a short historical window (at most 60 seconds), e.g. to see exactly how a price moved around an event. Requires the user's own Pyth Pro access token. If the user configured it in their MCP client, omit `access_token`; otherwise pass it as `access_token`, and if the tool reports it missing, ask the user for it. Symbols can be full names from get_symbols (e.g. 'Crypto.BTC/USD', 'Equity.US.AAPL/USD') or bare pairs like 'BTC/USD'. A bare pair resolves to the live spot feed when there is one, otherwise to the only remaining live match; `resolved_symbols` in the response shows what each input resolved to, and an ambiguous input returns an error listing the candidates. If both price_feed_ids and symbols are provided, only price_feed_ids are used. Rows are ordered by time and interleaved across feeds; a fixed_rate@200ms channel yields 5 rows per feed per second. When `has_more` is true, call again with `after` set to `next_cursor` and the same feeds, window and channel. Historical data is available from April 2025 onward; do not request times before that. Prices are integers with an exponent field: human-readable price = price * 10^exponent. Pre-computed display_* fields (display_price, display_bid, display_ask, display_confidence, display_ema_price, display_ema_confidence, display_funding_rate) apply the exponent for you. Price tools: get_latest_price for now, get_historical_price for one past instant, get_price_range for every update within up to 60 seconds, get_candlestick_data for OHLC bars over longer periods.
