The trading assistant

Ask, investigate, keep watch #

Describe what you want to understand: a contract, stock token, pool, wallet or market pattern. Ultron uses exact chain and contract identities, calls the available read tools, and returns evidence with source links and observation times.

The assistant follows three research lanes: meme or clip narrative, real product or infrastructure utility, and stock/ticker catalysts. A token does not need a viral video to be worth investigating. Volume, momentum and liquidity can justify a research alert without invented social proof.

Discovery is a sample #

DEX Screener and GeckoTerminal discovery uses bounded indexed feeds. Two-hour and calendar-day rankings use complete candles for up to ten candidate pools. These results are samples, never a claim to cover every coin on the chain. Paid visibility does not establish organic demand.

FOMO-first discovery is available when your connected agent supplies a recent FOMO snapshot. Ultron does not currently run a continuous FOMO board feed. Imported observations are explicitly labeled unverified and become stale.

What an answer should contain #

A useful answer identifies the token and network, says what was actually measured, links the source, notes the observation window and flags missing information. Market cap and fully diluted valuation are separate fields. Headlines are leads, not proof of a catalyst.

Public wallet labels and shared funders are clues, not proof that accounts have the same owner. Indexed prices are not execution guarantees. Ultron cannot establish that trading volume is organic or that a token is safe.

Keep the ping short #

Connect Ultron to your existing AI through MCP. Approved watch conditions create events in your AI through scheduled checks. Telegram and Discord are coming soon.

Watch this exact pair when hourly volume exceeds $100,000,
price change exceeds 8%, and liquidity stays above $50,000.
Show me the rule before activating it.

What the current data supports #

Available now: indexed token prices and liquidity, token-specific source windows, ETH and explicitly requested token balances, native-balance threshold alerts, price/volume/liquidity conditions, stock-token reference data and optional social/news research. Alerts appear in your AI through scheduled checks; an external AI client only pushes notifications if it supports that behavior.

Use get_market_history for complete five-minute or hourly pool candles, and compare_market_windows for equal-window volume comparisons. Rankings remain a bounded provider sample. Alchemy-backed wallet transfers require your key and method coverage; opposing flows are only candidate swaps. FIFO profit calculations require recorded USD cost basis and do not reconstruct an entire wallet automatically.

Context inside your AI #

Authenticated MCP calls retrieve relevant excerpts from maintained Robinhood research. Your agent combines dated case evidence with current tool results. The research library itself is not public, and old market outcomes are not current facts or guarantees.

Ask “What can you do?” for a short capability menu. get_trader_capabilities contains 50 example workflows and the source or setup each needs.