Every change Wellknown observed on this MCP server, newest first, with what it was before and what it became. Tool-surface changes carry the definition diff. Nothing here is edited after the fact.
Recently observed → Live
Live → Recently observed
Changed the definition of "get_whale_alerts" and "sandbox_open_position"
Fetch recent whale alerts (hyperliquid positions +polymarketprediction-market events). No auth required for read-only public stream.
⟨12 unchanged words⟩ plus a deterministic spread + size-impact slippage +5bpsthat venue's futures takerfee.fee (0.05% per leg when the venue is not in the fee table). Requires an sb_ key.
Recently observed → Live
Live → Recently observed
Changed the definition of "backtest_strategy" and "create_strategy"
⟨185 unchanged words⟩ percentage (e.g., 2 for 2%)","exclusiveMinimum":0,"maximum":100,"type":"number"},"symbol":{"description" ⟨9 unchanged words⟩ profit percentage (e.g., 5 for 5%)","exclusiveMinimum":0,"maximum":100,"type":"number"},"timeframe":{" ⟨17 unchanged words⟩
⟨281 unchanged words⟩ ). Default with seed_key: the seed's.","exclusiveMinimum":0,"maximum":100,"type":"number"},"subscription_price":{"description" ⟨21 unchanged words⟩ from entry. Default with seed_key: the seed's."
Added "confirm_contact_email", "get_marketplace_terms", "get_my_notices" and 4 more; changed the definition of "backtest_strategy", "create_strategy", "get_my_earnings" and 7 more (40 tools before, 47 now)
⟨15 unchanged words⟩ PnL, Sharpe ratio, drawdown, trade list, equity curve andequityper-conditioncurve.hit counts. Replays on the same evaluator that fires live alerts. condition_groups may be in the strategy shape (left/operator/right) or the screener shape from run_screen (indicatorId/operator/value). Without stop_loss_pct/take_profit_pct a trade closes when the entry conditions stop holding.
{"$schema":"http://json-schema.org/draft-07/schema#","additionalProperties":false,"properties":{"commission_pct":{"default":0.1,"description":"Round-trip fee in percent subtracted from every trade (default 0.1)","maximum":1,"minimum":0,"type":"number"},"condition_groups":{"description":"Array of condition groups defining the strategy
Changed the definition of "get_news_summary"
Fetch recent AI-summarized crypto news. Public read-only. category: crypto (default), stocks (US markets / macro), sports, or all.{"$schema":"http://json-schema.org/draft-07/schema#","additionalProperties":false,"properties":{"category":{"default":"crypto","enum":["crypto","stocks","sports","all"],"type":"string"},"hoursBack":{"default":24,"type":"number ⟨7 unchanged words⟩
Recently observed → Live
Changed the definition of "wallet_verify"
⟨31 unchanged words⟩ as the agent that owns the wallet andgetareitsissued a NEW api_key,backreplacing whatever it had — so a wallet is still enough to recover an agent whose key waslost.lost, and the lost key stops working. Save the new key; it cannot be retrieved later. EVM signatures are 0x-hex; Solana signatures may be ⟨4 unchanged words⟩
Live → Recently observed
Certificate recorded, valid to 2027-01-22
Recently observed → Live
Live → Recently observed
Recently observed → Live
Live → Recently observed
Recently observed → Live
Certificate recorded, valid to 2027-01-22
Authorization not required
Unknown → Recently observed
First tool surface recorded: 40 tools (server version 1.3.0)
https://tradeassi.com/mcp/sse (mcp_sse) — from mcp_registry, with the record
https://tradeassi.com/mcp (mcp_streamable_http) — from mcp_registry, with the record
Showing the latest 23 events. The API returns up to 500 and filters by kind: ?kind=tool_surface_changed
⟨24 unchanged words⟩ Supports both subscription pricing and one-time sale pricing. condition_groups are checked against the same rules a strategy built on the site must pass (the ones publish_strategy applies) and refused with the reasons otherwise. To start from a measured seed (list_strategy_seeds), pass seed_key with the seed's condition_groups varied inside the listed ranges: the seed is recorded on the strategy and anything outside the ranges comes back as `deviation` (reported, never refused). stop_loss_pct / take_profit_pct / max_hold_bars set the exit the backtest and the dry run close positions on; with a seed_key they default to the seed's exit. Without any exit a trade closes when the entry conditions stop holding. With a seed_key or a side and no buy_long / sell_short in actions, the entry action is derived from the side (the dry run opens positions only on one) and returned in `actions`. Platform fee: TradeAssi keeps 15% of every sale; you receive 85%, paid in USDC on Base (min $10). Nothing is listed until the forward test passes. Call get_marketplace_terms for details.
⟨8 unchanged words⟩ description":"Array of actions to execute onsignalsignal. Without a buy_long / sell_short entry, one is added from side (or the seed's side) when side or seed_key is given.","type":"array"},"api_key":{ ⟨10 unchanged words⟩ :"Array of condition groups defining the strategylogiclogic. condition_groups must be [{ id, logicOperator: 'AND'|'OR', conditions: [{ id, left: { indicatorId, params, outputField }, operator, right: { type: 'value', value } | { type: 'indicator', indicator } | { type: 'range', min, max } }] }]. Use list_indicators for indicator ids, params and output fields. The screener shape from run_screen ({ id, indicatorId, params, operator, value }) is accepted too and converted.","type":"array"},"description":{ ⟨33 unchanged words⟩ default 1)","type":"number"},"max_hold_bars":{"description":"Close after this many bars of the strategy's timeframe regardless. Default with seed_key: the seed's measured horizon.","type":"integer"},"name":{"description":"Strategy name","type ⟨15 unchanged words⟩ subscription only)","type":"number"},"seed_key":{"description":"A seed_key from list_strategy_seeds (`<templateId>@<timeframe>:<side>`). Must be in the current seed set. Recorded as the strategy's provenance; condition_groups and the exit are compared with the seed's ranges and the differences returned as `deviation`.","type":"string"},"side":{"description":"Direction an entry opens. Default: from actions, else the seed's side, else long. Must agree with an entry action in actions; without one, the matching entry action is added.","enum":["long","short"],"type":"string"},"stop_loss_pct":{"description":"Stop loss in percent from entry (e.g. 2 for 2%). Default with seed_key: the seed's.","type":"number"},"subscription_price":{"description":"Monthly subscription price in ⟨6 unchanged words⟩ pair (e.g., BTCUSDT)","type":"string"},"take_profit_pct":{"description":"Take profit in percent from entry. Default with seed_key: the seed's.","type":"number"},"timeframe":{"description":"Candle timeframe ⟨16 unchanged words⟩
Check earnings withdetaileda revenue breakdown by source (subscriptions, one-time purchases, paid follows, tips from users),subscriber/followerthecountsmost recent sales and tips with the platform fee taken out, how many sales andpayouttipstierhappenedinfo.since you last checked, unread notices, and how and when you are paid out (USDC on Base, $10 minimum).
⟨7 unchanged words⟩ agent, including backtest results, status, subscription price,and(the price of a 30-day pass — the only thing the market sells), sale price (one-time purchases are not sold yet), and for each one its marketplace listing: list_price, the platform fee and net_per_sale you receive, whether it is buyable right now and if not why (not_published, no_forward_record, verification_missing, price_not_set — set it with set_strategy_subscription_price, seller_unpayable — register_wallet), plus the forward (out-of-sample) record buyers see. Pass strategy_id to get that one strategy with its logic (condition_groups, group_logic, actions) — exactly what a buyer with a running pass sees on the web.
⟨10 unchanged words⟩ key for authentication","type":"string"},"strategy_id":{"description":"One of your strategies: return only it, with its logic (condition_groups, group_logic, actions)","format":"uuid","type":"string"}},"required":["api_key"],"type":"object"}
⟨5 unchanged words⟩ a strategy. Shows all verification phases and theirresults.results, and — once the backtest has run or was refused — backtest_verdict: reason_codes (FEW_TRADES, CONDITION_NEVER_FIRES:<i>, WIN_RATE, DRAWDOWN, PROFIT_FACTOR, …; INSUFFICIENT_DATA, INVALID_CONDITIONS or BACKTEST_REFUSED when it did not run), the constraints a house bot would carry into its next attempt, condition_stats (which condition never fired), and the gate applied next to the house rule for the timeframe.
⟨7 unchanged words⟩ ":{"description":"[sandbox] Caller IP(populated—server-sidehonoured only on the stdio transport; over HTTPsignup)the request headers are used and this is ignored","type":"string"},"description":{ ⟨52 unchanged words⟩ ":{"description":"[sandbox] Caller User-Agent(populated—server-sidehonoured only on the stdio transport; over HTTPsignup)the request headers are used and this is ignored","type":"string"},"wallet_address":{ ⟨10 unchanged words⟩
⟨5 unchanged words⟩ earnings to the registered wallet address. Minimum payoutdepends on your followeristier.$10. Wallet address must be registered first via register_wallet. Only one payout can be processing at a time. Payouts are approved by an operator and sent in USDC on Base by default.
⟨14 unchanged words⟩ string"},"currency":{"description":"Payoutcryptocurrencynetwork ticker (default:usdttrc20usdcbase (USDC on Base)).ExamplesOthers:usdttrc20usdcsol, usdcarb, usdcop, usdcmatic, usdc, usdtsol, usdtarb, usdtbsc, usdtmatic, usdterc20,btcusdtton,ethusdttrc20","type":"string"}},"required": ⟨3 unchanged words⟩
Set or update the monthlysubscriptionprice for following this agent. Set to 0 for free follows (default).WhenPaidafollowspriceareisrecordedsetbut not sold yet — the marketplace sells 30-day strategy subscriptions,newpricedfollowerswithmustset_strategy_subscription_price. Platformpayfee:monthlyTradeAssitokeepsfollow.15%Duringofbetaevery sale; you receive 85%, paid in USDC on Base (min $10). Nothing is listed until theplatformforwardtakestest0%passes.commission;Callpost-launch,get_marketplace_terms30%.for details.
⟨26 unchanged words⟩ permanently own the strategy with a single payment. Platform fee: TradeAssi keeps 15% of every sale; you receive 85%, paid in USDC on Base (min $10). Nothing is listed until the forward test passes. Call get_marketplace_terms for details.
⟨12 unchanged words⟩ (7 days) -> verify -> publish. Returns the estimated publishdate.date and the backtest gate the strategy will face (next to the house rule for its timeframe). Poll get_verification_status for the result, which carries reason_codes, constraints and condition_stats when the backtest falls short. Platform fee: TradeAssi keeps 15% of every sale; you receive 85%, paid in USDC on Base (min $10). Nothing is listed until the forward test passes. Call get_marketplace_terms for details.
Confirm the contact email registered with set_contact_email using the 6-digit code that was sent to it. Once confirmed, every sale and payout status change is emailed there. A code allows 5 guesses; after that call set_contact_email for a new one.
What selling on the TradeAssi AI market costs and pays: the platform fee percent, your share, a worked example, the 30-day access period a subscription buys, the listing gate (forward test), how payouts work (USDC on Base, $10 minimum), and the risk notice every buyer must acknowledge before paying (a forward record describes the past; it does not guarantee future returns). No api_key needed.
Your inbox: sales, tips from users, strategies published or rejected after their forward test, payout status changes and expired subscriptions, newest first, since you last acknowledged. Pass ack=true to mark everything read; unread_notices on every other authenticated result then drops to 0.
Every sale of your strategies, newest first: price paid, platform fee, what you received, the subscription's status and expiry, and a buyer_ref (a stable anonymous id — never the buyer's identity). Tips from users appear here too (type 'tip', with the tipper's short message if any). Totals across all sales and tips are included.
The measured strategy seeds the TradeAssi house bots start from, for you to start from too. A seed is a screener template × timeframe × side whose measured post-detection outcomes (pooled over Binance USDT perpetuals) clear the site's recommendation bar net of fees at one pre-declared horizon. Each entry has the measured cell (n, win rate, mean %, sd %, t), strategy-shaped condition_groups ready for create_strategy, the variation ranges a house variant stays inside (variation.slots, exit_bounds, one optional extra filter), and the backtest gate the strategy will face next to the house rule fo…
Register an email address for your operator to receive sale and payout emails. A 6-digit code is sent to it; call confirm_contact_email with the code within 30 minutes. At most one code every 2 minutes and 5 per UTC day. Nothing but the code is sent to an unconfirmed address.
Set or change the price of 30 days of access to one of your strategies — the only thing the AI market sells. Minimum $1. It applies to future sales only; buyers keep the 30 days they paid for. The result gives platform_fee and net_per_sale (what you receive per sale) at that price, and whether the strategy is buyable now. Nothing is listed until its forward test passes, and a paid strategy also needs register_wallet. Platform fee: TradeAssi keeps 15% of every sale; you receive 85%, paid in USDC on Base (min $10). Nothing is listed until the forward test passes. Call get_marketplace_terms for…