> ## Documentation Index > Fetch the complete documentation index at: https://partners.centaur.io/docs/llms.txt > Use this file to discover all available pages before exploring further. # Prompts to Get Started > Starter prompts that steer MCP clients toward supported Centaur reads. Use these prompts after Centaur MCP is connected. They are written to encourage evidence-backed reads instead of unsupported prediction or ranking claims. ## Daily Feed Intelligence > Use Centaur `list_feed` to show the latest curated source-message groups. For each group, name the trader, quote the source text briefly, and list its events with asset, direction, and event type. > Use Centaur to summarize the dominant market themes from Generated Aggregate Narrative Summaries over the last 24 hours. Use explicit UTC bounds and say if the result is partial. > Use Centaur to find recent trade events from the last 24 hours, skipping assumed or auto-generated system events. Summarize the most notable source-backed trades. > Use Centaur to show the last 10 trader-sourced events from the visible universe. Include trader, asset, direction, event type, and source-message IDs when available. ## Current Positioning > Use Centaur open-position reads to summarize current open-position skew across major assets. Use counts and percentages where available. > Use Centaur to resolve BTC, then list current open positions on BTC with trader names, direction, entry price, and current return when available. ## Rankings and Activity Trends > Use Centaur to rank the most active traders of the last 7 days with `rank_traders`. Include event and position counts and direction bias. > Use Centaur to rank traders by `7D` win rate over positions opened in the last 30 days with `rank_traders`. Qualify thin samples using the returned positions count. > Use Centaur to show daily message and event volume for the last 7 days with `summarize_message_activity`, grouped by trader. Name the most active traders. ## Trader Performance > Use Centaur to resolve candidate traders, then compare their `7D` time-based performance over positions opened in the last 30 days. > Use Centaur to list traders with strong win rate in a defined sample. State the sample bounds, performance window, and any missing IDs or partial pages. > Use Centaur to compare a trader's `1D`, `7D`, and `30D` time-based performance. Explain unavailable windows by status. ## Asset Deep Dive > Use Centaur to resolve BTC, then analyze visible BTC activity: current open-position skew, recent source-backed events, and top trader IDs from asset stats. > Use Centaur to compare current positioning on ETH with generated aggregate summary narratives from the same UTC day. Keep positioning and narrative claims separate. ## Trader Profile > Use Centaur to resolve Crypto Capo, then summarize their visible profile: asset focus, direction bias, time-based performance, current open positions, and recent source messages. > Use Centaur to hydrate the source message for this event ID path: first inspect the event, then use `positionId` for performance and `messageId` for source context. ## Strategy Research > Use Centaur to find assets with visible open-position skew in one direction, then identify which traders are currently positioned there. Do not treat this as a prediction. > Use Centaur to find open positions whose current return is between 5% and 10%. Qualify the answer as current open-position state, not completed trade performance. > Use Centaur to compare assets where multiple visible traders are currently positioned in the same direction. Include counts and note if pagination means the result is partial. ## Tips * Start with `list_traders` or `list_assets` when you do not know IDs. * Use `list_feed` for presentation-ready source-message groups with curated events instead of joining separate event and message reads. * Use open positions for current positioning. * Use stats and position history for performance. * Use generated summaries for narratives, themes, drivers, and risks. * Use `rank_traders` for trader rankings and `summarize_message_activity` for count and trend questions instead of paging raw events or messages.