Official MCP RegistryListed
Forex-GPT MCP Server
AI-powered forex and crypto trading analysis with real-time data across 127+ instruments.
First seen 2 Oct 2026. Evidence as of 2 Oct 2026.
69
Tools
From an anonymous probe
1
Source listings
Each with its own history
0
Recorded changes
Since first seen
Tools
| Tool | Description | Behaviour |
|---|---|---|
| add_to_watchlist | Add a new asset to your persistent watchlist. Args: asset: Asset symbol (e.g. 'EUR_USD', 'BTC_USD') | Changes data |
| agent_approve | Approve a pending agent action (trade entry, exit, modification). Args: action_id: The pending action ID to approve adjust_sl: Optional adjusted stop-loss price adjust_tp: Optional adjusted take-profit price adjust_size: Optional adjusted position size (units) Returns: Result of the executed action Example: "Approve action xyz789" or "Approve the pending EUR/USD trade" | Read-only |
| agent_history | Get historical performance and actions for an agent. Args: agent_id: Specific agent, or omit for all agents days: Number of days of history (1-90). Default: 7 include_cycles: Include detailed scan cycle logs include_trades: Include trade history. Default: True limit: Max trades per agent (the Pro Terminal dashboard sends this; unknown args previously made the whole call fail) Returns: Historical performance data, trade log, and statistics Example: "Show me my agent's trading history for the past week" | Read-only |
| agent_link_telegram | Link your Telegram account to receive agent notifications. To get your Telegram chat ID: 1. Open Telegram and search for @userinfobot 2. Start the bot and it will show your ID 3. Copy the number and use it here Args: telegram_chat_id: Your Telegram chat ID (a number like "123456789") Returns: Confirmation of successful linking Example: "Link my Telegram with chat ID 123456789" | Read-only |
| agent_list_strategies | List all available trading strategies with descriptions. Returns information about each strategy including: - Recommended timeframes and instruments - Typical hold times - Risk profile - Minimum scan interval Example: "What trading strategies can the agent use?" | Read-only |
| agent_modify | Modify a running agent's parameters. Args: agent_id: The agent ID to modify risk_per_trade: New risk per trade percentage (0.1-5.0) max_positions: New maximum concurrent positions (1-10) control_level: New control level ("manual", "supervised", "semi_auto", "full_auto") scan_interval_minutes: New scan interval in minutes instruments: Replace instrument list entirely add_instruments: Add instruments to existing list remove_instruments: Remove instruments from list custom_rules: Update custom instructions pause: True to pause agent, False to resume min_score: Minimum sentiment score for scan filtering (0-100) sentiment_threshold: (HFT) Minimum abs(score) to trigger signal (50-100) sl_pips: (HFT) Stop loss in pips (1-50) tp_multiplier: (HFT) Take profit multiplier on SL distance (1.0-5.0) exclude_categories: Asset categories to exclude from scanning/trading. Options: "forex", "crypto", "indices", "commodities", "metals", "bonds". Pass empty list [] to clear exclusions. exclude_instruments: List of specific instruments to exclude from trading, e.g. ["USD_TRY", "EUR_TRY"]. Pass empty list [] to clear exclusions. allow_closed_markets: If True, generate signals for closed markets using limit orders instead of market orders. Default: False (skip closed markets). auto_scan: Enable/disable dynamic scanning. When True with empty instruments, agent rescans market each cycle to find fresh opportunities. Returns: Updated agent configuration Example: "Change my agent to 0.5% risk" or "Add GBP/USD to the agent" | Read-only |
| agent_notification_settings | View your notification settings and linked accounts. Returns: Your current notification preferences and linked services Example: "Show my notification settings" | Read-only |
| agent_reject | Reject a pending agent action. Args: action_id: The pending action ID to reject reason: Optional reason for rejection (helps agent learn) pause_agent: If True, pause the agent after rejection Returns: Confirmation of rejection Example: "Reject that trade, the spread is too high right now" | Read-only |
| agent_start | Start an autonomous trading agent with specified strategy. This creates a new agent that will scan markets, analyze opportunities, and trade (or request approval) based on your configuration. Args: strategy: Strategy type - "swing", "scalper", "news_trader", "trend_follower", "breakout", "mean_reversion", "hft" instruments: List of instruments to trade, e.g. ["EUR_USD", "XAU_USD"]. Optional if auto_scan=True. auto_scan: If True, automatically scan market to find high-scoring opportunities. Default: False scan_min_score: Minimum score (0-100) for auto-scanned instruments. Default: 70 scan_limit: Maximum number of instruments from scan results. Default: 10 exclude_instruments: List of instruments to exclude from trading, e.g. ["BTC_USD", "ETH_USD"] timeframes: Timeframes to analyze, e.g. ["H1", "H4", "D"]. Defaults to strategy recommendation. risk_per_trade: Maximum risk per trade as % of account (0.5-5.0). Default: 1.0 max_positions: Maximum concurrent open positions (1-10). Default: 3 control_level: Autonomy level. Options: - "manual": Analysis only, no trading - "supervised": Agent proposes trades, you approve each one - "semi_auto": Auto-execute small trades, approve larger ones - "full_auto": Fully autonomous within risk limits scan_interval_minutes: How often to scan market (5-1440). Default: 60 notifications: Channels for alerts - ["telegram", "email", "push"]. Default: ["telegram"] custom_rules: Optional custom instructions for the agent min_score: Minimum sentiment score for scan filtering (0-100). Overrides strategy default. sentiment_threshold: (HFT) Minimum abs(score) to trigger signal (50-100). Default: 80 sl_pips: (HFT) Stop loss in pips (1-50). Default: 10 tp_multiplier: (HFT) Take profit multiplier on SL distance (1.0-5.0). Default: 1.5 exclude_categories: Asset categories to exclude from scanning/trading. Options: "forex", "crypto", "indices", "commodities", "metals", "bonds". e.g. ["crypto", "bonds"] to skip all crypto and bond instruments. allow_closed_markets: If True, generate signals for closed markets using limit orders (executed when market reopens). Default: False (skip closed markets). cooldown_enabled: If True, enforce cooldown period after trading a pair. Default: True. cooldown_hours: Hours to wait before re-trading same instrument. Default: 4. allow_wide_spreads: If True, trade even when spread is abnormally wide (3x+ typical). Default: False. Useful for exotic pairs with naturally wide spreads. Returns: Agent details including ID, status, and next scan time Examples: "Start a scalper agent, auto-scan for opportunities with score > 80" "Start a swing agent on EUR/USD and Gold, exclude crypto" "Start an HFT agent on EUR/USD with 5 pip stop loss and 2x take profit" | Read-only |
| agent_status | Get status of one or all your agents. Args: agent_id: Specific agent ID, or omit for all agents Returns: Agent status including P&L, open positions, pending approvals Example: "What is my agent doing?" or "Show status of agent abc123" | Read-only |
| agent_stop | Stop a running agent. Args: agent_id: The agent ID to stop close_positions: If True, close all positions opened by this agent cancel_pending: If True, cancel all pending orders from this agent Returns: Confirmation of agent stop and any actions taken Example: "Stop agent abc123 and close all its positions" | Read-only |
| analyze_asset | Run a one-shot technical analysis for a SINGLE asset on a SINGLE timeframe. Use this when the user asks about ONE specific asset by name, e.g.: - "analyze EUR_USD" - "give me your read on BTC on the 4-hour" - "what's the signal on gold?" - "deep analysis on SOL/USDC daily" For multi-asset surveys ("scan the market", "find opportunities", "what looks bullish today") use scan_market INSTEAD — that scans the full universe and returns top setups by sentiment score. Calling analyze_asset for a market-scan query is incorrect and wastes tokens; the user wants a ranked list, not a deep dive on one asset they didn't even name. Args: asset: Asset symbol (e.g., EUR_USD, XAU_USD, BTC, ETH) timeframe: Timeframe for analysis (D, H4, H1, M30, M15, M5, M1) priceType: Price type - B (bid), A (ask), or M (mid) timeoutSec: Timeout in seconds (5-120) language: Optional. The language for the written analysis and the spoken narration — "es", "pt", "zh", "ja". Omit it and nothing changes: the prose follows the language stored on the user's profile, which is what every existing client gets today. Pass it and it wins, because the caller knows what is on screen now while a stored preference read from another process can be a cache interval behind. Numbers, instrument symbols and every structured field stay English either way. Returns: Analysis results with status, signals, and recommendations | Read-only |
| analyze_timeframes | Analyze a single asset across multiple timeframes simultaneously. Args: asset: Asset symbol (e.g., EUR_USD) timeframes: List of timeframes (e.g., ["D", "H4", "H1"]) priceType: B (bid), A (ask), or M (mid) timeoutSec: Overall timeout in seconds (30-300) Returns: Multi-timeframe analysis results | Read-only |
| analyze_watchlist | Analyze multiple assets from the user's SAVED watchlist in parallel. Args: timeframe: Timeframe (D, H4, H1, etc.) priceType: B (bid), A (ask), or M (mid) timeoutSec: Overall timeout in seconds (30-300) (optional) timeframes: list of timeframes (timeframes mode) – if provided, supersedes timeframe (optional) max_assets: limit how many watchlist assets to analyze (default: all) Returns: Results for all watchlist assets with analysis data | Read-only |
| backtest_history | List recent backtest runs, optionally filtered by instrument. Args: instrument: Filter by instrument (optional) limit: Max results (1-50). Default: 20 Returns: List of recent backtest runs with summary stats Example: "Show my backtest history" or "Show backtests for XAU_USD" | Read-only |
| backtest_list_saved | List saved parameter presets, optionally filtered by instrument. Args: instrument: Filter by instrument (optional) Returns: List of saved parameter sets Example: "Show my saved backtest params" or "Show saved params for EUR_USD" | Read-only |
| backtest_optimize | Grid search optimization to find the best parameter set for an instrument. Tests all combinations of sentiment_threshold, SL distance, and tp_multiplier and ranks results by the chosen metric. On forex/OANDA the SL grid is sl_pips_range (pips); on Hyperliquid, where pips are meaningless for crypto perps, it is sl_pct_range (% of price). Args: instrument: Instrument (e.g. EUR_USD, XAU_USD; BTC on Hyperliquid) timeframe: Candle granularity. Default: H1 count: Historical candles to fetch (100-5000). Default: 5000 threshold_range: Comma-separated sentiment thresholds to test. Default: "70,75,80,85,90,95" sl_pips_range: Comma-separated SL values in pips (forex). Default: "5,8,10,15,20" tp_mult_range: Comma-separated TP multipliers. Default: "1.0,1.5,2.0,2.5,3.0" rank_by: Metric to rank by: "sharpe", "pnl", "profit_factor", "win_rate". Default: "sharpe" initial_balance: Starting balance. Default: 10000 lookback: Candles lookback for indicators. Default: 120 spread_pips: Estimated spread in pips (forex). Default: 2.0 sl_pct_range: Comma-separated SL values as % of price (Hyperliquid). Default: "0.5,1.0,1.5,2.0,3.0" spread_pct: Estimated spread+slippage as % of price (Hyperliquid). Default: 0.05 Returns: Best parameters and top 50 results ranked by chosen metric Example: "Optimize EUR_USD on H4 and find the best threshold and SL" | Read-only |
| backtest_run | Run a backtest on historical data using sentiment-based signals. Uses the same deterministic sentiment scoring as the live scanner (Bollinger Bands + SMA alignment) to generate buy/sell signals when abs(score) >= threshold. Distance mode is automatic: forex/OANDA instruments use pips (sl_pips/spread_pips); on Hyperliquid, where pips are meaningless for crypto perps, SL and spread are a percent of price (sl_pct/spread_pct) and trade "pips" figures are percent moves. Args: instrument: Instrument (e.g. EUR_USD, XAU_USD; BTC on Hyperliquid) timeframe: Candle granularity — M1, M5, M15, M30, H1, H2, H4, H6, H8, H12, D or W (aliases like 1h/4h/daily accepted; monthly is not available). Default: H1 count: Number of historical candles (100-5000). Default: 5000 sentiment_threshold: Minimum abs(score) to trigger signal (50-100). Default: 80 sl_pips: Stop loss in pips (forex). Default: 10 tp_multiplier: Take profit as multiple of SL. Default: 1.5 risk_per_trade: Risk per trade as % of balance. Default: 1.0 max_positions: Max concurrent positions. Default: 1 spread_pips: Estimated spread in pips (forex). Default: 2.0 lookback: Candles lookback for indicators. Default: 120 initial_balance: Starting balance for simulation. Default: 10000 sl_pct: Stop loss as % of price (Hyperliquid/crypto). Default: 1.0 spread_pct: Estimated spread+slippage as % of price (Hyperliquid). Default: 0.05 Returns: Backtest results with stats, equity curve, and trade list Example: "Backtest EUR_USD on H4 with 90 threshold" | Read-only |
| backtest_save_params | Save a backtest's parameters as a reusable preset. Args: backtest_id: The backtest run ID to save params from name: Name for this parameter set (e.g. "EUR_USD H4 aggressive") Returns: Saved parameter set details Example: "Save those backtest params as 'EUR_USD H4 optimal'" | Read-only |
| cancel_pending_order | Cancel a pending (unfilled) Limit or Stop order. Args: order_id: The OANDA ID of the order to cancel. | Destructive |
| chart_chat_proxy | Proxy to the Flask /api/chart-chat endpoint using the caller's bearer token. Keeps token accounting/auth consistent with the web app while exposing chat via MCP. | Read-only |
| close_position | Close position for an instrument. | Destructive |
| close_trade | Close a specific trade (fully or partially). | Destructive |
| delete_oanda_credentials | Remove/Delete your stored OANDA API Key and Account ID. Use this if you want to clear your credentials or switch accounts. | Destructive |
| economic_events | Fetch upcoming economic events via the TradingView proxy source. Args: fromDays: Start of date range (days from now, 0-90) toDays: End of date range (days from now, 0-90) Returns: List of economic events with importance, country, and timing | Read-only |
| fetch | Retrieve complete forex analysis by ID. Args: id: Analysis ID in format "ASSET_TIMEFRAME" (e.g., "EUR_USD_D", "XAU_USD_H4") Returns: Full analysis document with technical indicators, signals, and recommendations | Read-only |
| get_account_history | Get account transaction history (e.g. closed trades, order fills). Useful for finding past trades, reporting, or auditing. Args: from_time: Start time (RFC3339 format, e.g. "2023-05-01T00:00:00Z"). to_time: End time (RFC3339). type_filter: Comma-separated types (e.g. "ORDER_FILL,TRADE_CLOSE"). To see closed trades, use "TRADE_CLOSE". | Read-only |
| get_blog_articles | Get recent articles from the Forex-GPT blog feed. Args: max_results: Maximum number of articles to return (1-50, default: 10) category: Optional category slug to filter by. When unset, the tool falls back to a broker-aware default: - hyperliquid containers → "daily-crypto-hype" (ID 24) so HL users see only crypto-focused content, not the forex daily AI analysis stream. - Anything else → site-wide feed. Pass an explicit slug (e.g. "daily-deep-dive", "the-weekly-open") to override the default. Returns: List of recent blog articles with titles, URLs, summaries, images, and publication dates | Read-only |
| get_chart_data | Retrieve OHLC (candlestick) data for charting. Args: asset: Asset symbol (e.g., EUR_USD) timeframe: Timeframe (D, H4, H1, etc.) count: Number of candles (50-5000) Returns: Candlestick data with timestamps, open, high, low, close prices | Read-only |
| get_crypto_news | Aggregated crypto-news feed from quality free RSS sources (CoinDesk, Decrypt, Blockworks, The Defiant). Sorted newest-first, deduped by URL across sources. Args: max_results: 1-50, default 25. ticker: Optional symbol filter (e.g. "BTC", "ETH", "SOL", "HYPE"). When set, only articles mentioning the ticker (case-insensitive substring match in title OR summary) are returned. Pass "all" or empty to disable filtering. Returns: {status, count, ticker, articles: [{title, url, summary, published, source, image?}]} | Read-only |
| get_instrument_cache_status | Read-only | |
| get_instrument_specs | Read-only | |
| get_live_heatmap | Generates a real-time Sentiment Heatmap snapshot. Behavior: 1. If 'assets' are provided, scans those specific assets. 2. If 'assets' not provided, scans the user's saved Watchlist. 3. If Watchlist is empty, scans a default list of popular global assets. Args: assets: Optional list of symbols. timeframe: Granularity (e.g. 'H1', 'D'). | Read-only |
| get_market_categories | Read-only | |
| get_market_hours | Read-only | |
| get_market_movers | Top gainers, top losers and (Hyperliquid) top 24h volume across the active broker's whole universe, ranked by daily price change. No Sentiment Score is computed and no tokens are charged — use scan_market for score rankings. Hyperliquid: rolling 24h change from the exchange's own reference price, plus 24h notional volume. OANDA: current daily candle vs the prior close. Args: limit: rows per list (1-50, default 25). liquid_only: Hyperliquid only — rank gainers/losers within the 100 most-traded markets by 24h notional so thin markets don't dominate. timeframe: OANDA only — candle granularity for the change (default "D"). | Read-only |
| get_open_positions | List open positions (Net Exposure per instrument). Useful to see total Long/Short units and P&L per pair. | Read-only |
| get_open_trades | List open trades. | Read-only |
| get_order_history | Get OANDA order history (FILLED, CANCELLED, TRIGGERED, etc.). state: "ALL", "PENDING", "FILLED", "TRIGGERED", "CANCELLED". count: Max 500. | Read-only |
| get_pending_orders | List all PENDING orders (Limits, Stops, Entry Orders). These are orders waiting to be filled. | Read-only |
| get_performance_summary | Trading-performance aggregates: realized P&L net of fees, total fees paid, perp funding paid/received, fill count, plus a daily breakdown for the chart. Args: window: "7d" | "30d" | "90d" | "ytd" | "all" (default). Filters fills + funding by their HL timestamp. Lifetime totals are always returned alongside the windowed slice. | Read-only |
| get_price_stream_url | Return a short-lived URL for the raw price stream (SSE) for given instruments. The URL expires after ~10 minutes. | Read-only |
| get_sentiment_stream_url | Return a short-lived URL for the sentiment stream (SSE) for given instruments and granularity. The URL expires after ~10 minutes. | Read-only |
| get_server_time | Return the server's current UTC timestamp (ISO 8601). | Read-only |
| get_spreads | Checks the current liquidity and trading cost (spread) for an asset. Useful for checking if a market is too expensive to trade right now. Args: asset: The asset symbol (e.g., "EUR_USD", "XAU_USD") Returns: Current spread, average spread, and cost in basis points. | Read-only |
| get_token_balances | Return the current user's token balances from Auth0 app_metadata. Uses _safe_profile() so that the same values shown inside AI analysis (token_quota, purchased_tokens, total_token_usage, last_reset, email, user_id) are returned in a simple JSON structure. | Read-only |
| get_trade_details | Get full details of a specific trade (Open or Closed). Useful to investigate a trade that disappeared or to see its full history. | Read-only |
| get_trading_account | Get trading account summary (balance, margin). | Read-only |
| get_watchlist | View the assets the user has SAVED to their persistent watchlist. Use this when the user asks about their watchlist, e.g.: - "show my watchlist" - "what's on my watchlist" - "list my watchlist" - "what assets am I watching" This is NOT the same as open positions / trades. The watchlist is a saved list of asset SYMBOLS the user follows; it has no balance, P&L, units, or entry-price data. For open positions use get_open_positions; for open trades use get_open_trades. Do NOT call get_open_positions or get_open_trades for watchlist queries — you'd return position data that doesn't match what the user asked. Returns: {count, assets, status} — list of saved asset symbols and count. | Read-only |
| health | Comprehensive health and status check including optional upstream dependencies. Args: checkOpenAI: Verify OpenAI API connectivity checkTradingView: Verify TradingView API connectivity checkOANDA: Verify OANDA API connectivity Returns: Health status with environment info and dependency checks | Read-only |
| help_tutorial | Master tutorial for MCP tools: usage tips, key parameters, and suggested flows. | Read-only |
| is_market_open | Read-only | |
| list_assets | Return supported asset symbols for the active broker. For OANDA containers: returns the OANDA universe (~120 instruments). For Hyperliquid containers: returns the Hyperliquid perp universe (~230 markets) with per-asset metadata (current price, max leverage, funding rate, open interest, 24h volume, product_type, sector, is_pre_launch, is_hip3). When ``include_spot=True`` (default), the spot universe from ``/info{spotMeta}`` is appended — instruments tagged ``product_type="spot"`` use ``_USDC`` suffix so they don't collide with perp canonical names. Optional: • filter: substring (case-insensitive) to narrow the list • asset: specific symbol to test membership • include_spot: append HL spot markets (HL only; default True) | Read-only |
| purchase_tokens_x402 | Purchase premium tokens via X402 crypto payment. Pay $20 USDC to receive 500,000 premium tokens (non-expiring). These tokens can be used to access all ForexGPT tools and services. Args: payment_proof: X402 payment proof containing signature and payment details. Must include: amount, token, network, nonce, v, r, s, from, to, validAfter, validBefore Returns: Success message with new token balance or payment required error. Example payment_proof: { "type": "eip3009", "amount": "20000000", # $20 USDC (6 decimals) "token": "USDC", "network": "eip155:84532", # Base Sepolia "from": "0x...", # User's wallet "to": "0x...", # Recipient (ForexGPT treasury) "nonce": "0x...", "validAfter": 1234567890, "validBefore": 1234567890, "v": 27, "r": "0x...", "s": "0x..." } | Read-only |
| purchase_tokens_x402_solana | Purchase premium tokens via X402 payment on Solana. Pay $20 USDC on Solana (devnet or mainnet) to receive 500,000 premium tokens. Credits go to the same purchased_tokens balance used by the Stripe and EVM rails — one balance, three rails. Two accepted proof shapes: signed_authorization (off-chain Ed25519 sig over a canonical message): { "type": "signed_authorization", "network": "solana:devnet", "token": "USDC", "from": "<signer pubkey base58>", "to": "<treasury pubkey base58>", "amount": "20000000", "nonce": "<random hex>", "valid_before": <unix seconds>, "signature": "<base58 ed25519 signature>" } settled_tx (confirmed SPL USDC transfer, fetched and diffed via RPC): { "type": "settled_tx", "network": "solana:devnet", "token": "USDC", "from": "<signer pubkey base58>", "to": "<treasury pubkey base58>", "amount": "20000000", "nonce": "<idempotency key>", "tx_signature": "<base58 transaction signature>" } | Read-only |
| refresh_instrument_data | Manually refresh instrument data from OANDA API. Updates financing rates, margin requirements, and other parameters. Requires: User must have saved OANDA credentials via save_oanda_credentials. Returns: - success: Boolean indicating if refresh succeeded - message: Status message - count: Number of instruments refreshed - refreshedAt: Timestamp of refresh Note: This fetches fresh data from OANDA's practice environment. The cache is automatically refreshed on server startup if data is >24 hours old. | Changes data |
| remove_from_watchlist | Remove an asset from your persistent watchlist. Args: asset: Asset symbol to remove | Changes data |
| replace_pending_order | Replace (Modify) an existing pending order. Basically cancels the old one and places a new one atomically. Args: order_id: ID of the order to replace. type: 'LIMIT' or 'STOP'. client_nonce: Optional. 32-char hex idempotency token for the new order. Pass the same value from a prior call's response to retry safely. Server mints one if absent. (All other args same as placement tools) | Destructive |
| research_briefing | Generate a short AI briefing (one paragraph + three bullets) from the terminal's current research context: watchlist Sentiment Scores, running agents and their theses, open positions, upcoming events / headlines and the latest research article. Charged to the user's tokens like any AI analysis (1 token = 1 LLM token); refused when the balance can't cover it. Args: context: JSON assembled by the terminal — keys: pulse, agents, positions, events, headlines, article, network, broker. | Read-only |
| save_broker_credentials | Save broker credentials. Supports multiple brokers. Args: broker: The broker name - "oanda", "etoro", or "hyperliquid" api_key: Your broker API key (OANDA/eToro). NOT used for Hyperliquid (see below). account_id: Your broker account ID (required for OANDA) user_key: Your eToro user key (required for eToro) demo_mode: Use demo/virtual portfolio (eToro only, default True) Note for Hyperliquid users: Hyperliquid does not use API keys. Instead you connect a self-custody Web3 wallet and authorize a trading agent via an on-chain EIP-712 signature. This MCP tool cannot perform that signature — it requires a wallet popup in the FE. Calling this tool with broker="hyperliquid" returns instructions instead of saving anything. Use the FE flow at Settings → Hyperliquid → "Connect Wallet" + "Authorize Trading". | Destructive |
| save_oanda_credentials | Save your OANDA API Key and Account ID securely. These are required before you can trade. | Destructive |
| scan_market | Scans the FULL market universe (Forex, Crypto, Metals, Indices) and returns the top trading opportunities ranked by sentiment score. Use this when the user asks for a MULTI-ASSET survey or anything that implies scanning / finding opportunities across the market. Example user phrasings that should trigger scan_market: - "scan the market" - "scan market on daily for score of 85" - "what's bullish on the H4?" - "find me strong setups" - "show top opportunities" - "what looks bearish right now?" - "any crypto setups?" For deep analysis of a SINGLE specific asset the user names by symbol, use analyze_asset instead. Do NOT call analyze_asset on a single asset when the user asked for a scan — that produces one deep dive instead of the ranked list they wanted. Common arg parsing from user phrasings: - "score of 85" / "min score 85" / "score >= 85" → min_score=85 - "on daily" / "on D" / "1d" → timeframes=["D"] - "on 4-hour" / "H4" / "4h" → timeframes=["H4"] - "top 10" / "top 5" → limit=10 / limit=5 - "bullish only" → sentiment="bullish" - "crypto" / "stocks" → include_categories=["crypto"] / ["stocks"] Args: timeframes: List of timeframes to scan (e.g. ["H1", "H4", "D"]). Default is ["H8"]. min_score: Minimum absolute sentiment score (0-100) to include. Default 80 (Bullish/Bearish). limit: Maximum number of results to return. Default 6. include_categories: Optional list of category names to keep. Each name matches the engine's `sector` (crypto/stocks/indices/commodities/fx) OR `product_type` (perp/spot) OR special tokens "hip3"/"pre-launch". Case-insensitive. Empty/None = include everything. exclude_categories: Optional list of category names to exclude. Applied AFTER include filter. Same vocabulary. universe_scope: "top25" (default; rank by 24h volume then keep top 25) or "all" (the 100 most-traded markets by 24h volume — a hard cap that keeps scan time and the shared exchange rate budget bounded; it is NOT the whole ~500-market universe). HL only — OANDA always uses its static universe regardless. min_volume_24h: Drop markets whose volume_24h is below this floor. HL only; OANDA universe has no volume metadata to filter on. Returns: A list of the top opportunities ranked by sentiment strength. | Read-only |
| search | Search forex analysis results by asset name, timeframe, or keywords. Returns a list of available analyses matching the search criteria. Args: query: Search query (e.g., "EUR_USD", "daily analysis", "gold", "bitcoin") Returns: Search results with IDs for fetching full analysis | Read-only |
| trade_limit_order | Place a Limit Order. price: The limit price to trigger the order. take_profit: ABSOLUTE PRICE (e.g. 1.1250). stop_loss: ABSOLUTE PRICE (e.g. 1.1100). take_profit_distance: RELATIVE PIPS (e.g. 0.0050). stop_loss_distance: RELATIVE PIPS (e.g. 0.0020). trailing_stop_distance: RELATIVE DISTANCE for trailing stop. guaranteed_stop_loss: True for GSLO. expiry: ISO 8601 string (e.g. "2023-12-31T23:59:00Z"). Defaults to GTC if omitted. client_nonce: Optional. 32-char hex idempotency token. reduce_only: HL only. Order may only close existing position size. tif: HL only. "Gtc" (default), "Alo" (post-only), or "Ioc". payment_proof: Optional. HL mainnet only — pass {"confirmed": true} to acknowledge orders whose notional exceeds HYPERLIQUID_MAINNET_CONFIRM_USD ($500 default). | Destructive |
| trade_market_order | Place a Market Order. units: Positive for Buy, Negative for Sell take_profit: ABSOLUTE PRICE (e.g. 1.1250). stop_loss: ABSOLUTE PRICE (e.g. 1.1100). take_profit_distance: RELATIVE PIPS (e.g. 0.0050). Use this OR absolute price. stop_loss_distance: RELATIVE PIPS (e.g. 0.0020). Use this OR absolute price. trailing_stop_distance: RELATIVE DISTANCE for trailing stop (e.g. 0.0020). guaranteed_stop_loss: True to use GSLO (Premium, requires specific SL distance rule). slippage: Hyperliquid-only — fraction (0–0.5) capping IOC price band. Ignored by OANDA/eToro. client_nonce: Optional. 32-char hex idempotency token. Pass the same value from a prior call's response.client_nonce to retry the exact same order without risking a duplicate fill. Server mints one if absent. payment_proof: Optional. HL mainnet only — pass {"confirmed": true} to acknowledge orders whose notional exceeds HYPERLIQUID_MAINNET_CONFIRM_USD ($500 default). UI shows a confirmation modal first, then re-fires with this set. | Destructive |
| trade_modify_trade | Modify an existing trade's Stop Loss and Take Profit. You can use absolute prices OR relative distances. Only provided fields will be updated. Args: trade_id: The ID of the open trade to modify. take_profit: New ABSOLUTE price. stop_loss: New ABSOLUTE price. take_profit_distance: New RELATIVE distance (e.g. 0.0050). stop_loss_distance: New RELATIVE distance (e.g. 0.0020). | Destructive |
| trade_stop_order | Place a Stop Order (Momentum/Breakout Entry). Use this to Buy ABOVE market price or Sell BELOW market price. Args: instrument: Asset symbol (e.g. 'EUR_USD') units: Positive for Buy, Negative for Sell price: The STOP price to trigger the order. expiry: ISO 8601 string (e.g. "2023-12-31T23:59:00Z"). Defaults to GTC if omitted. take_profit: ABSOLUTE PRICE. stop_loss: ABSOLUTE PRICE. take_profit_distance: RELATIVE DISTANCE (e.g. 0.0050). stop_loss_distance: RELATIVE DISTANCE (e.g. 0.0020). trailing_stop_distance: RELATIVE DISTANCE for trailing stop. guaranteed_stop_loss: True for GSLO. client_nonce: Optional. 32-char hex idempotency token. Pass the same value from a prior call's response.client_nonce to retry without risking a duplicate order. Server mints one if absent. | Destructive |
| x402_evm_config | Return the public EVM x402 config: recipient address, supported networks (CAIP-2 keys → USDC contract + RPC URL + chain name), and the token package. Pairs with x402_solana_config so the WalletPaymentModal can render a Solana | EVM tab switcher and let the user pay from whichever wallet they have. No auth — every value is broadcast on-chain at payment time anyway. | Read-only |
| x402_solana_config | Return the public Solana x402 config: recipient pubkey, supported networks (with USDC mints + RPC URLs), and the token package. No authentication required — every value here is broadcast publicly on-chain at payment time anyway. The frontend calls this on the payment modal's mount so the build doesn't need its own VITE_SOLANA_* env vars; one source of truth lives on the MCP container. | Read-only |
Change history
No changes since the first observation. The first snapshot is the baseline.
| Source | Listing | First seen | Last seen | Versions |
|---|---|---|---|---|
| Official MCP Registry | io.github.hatgit/forex-gpt-mcp | 2 Oct 2026 | 2 Oct 2026 | 1 |