free prompt kit · edgeful API

10 trading dashboards to build with AI agents.

copy-paste 10 prompts into Claude, Cursor, or any LLM connected to the edgeful API. each one pulls years of probability data, stacks the reports, does the math, and hands back a polished dashboard. zero code written.

why this exists

the edgeful UI answers one question at a time.

each report has depth, but you see one data point per screen. the real edges hide where two or three reports overlap, and that's exactly the math an agent can do for you.

the old way

view one report at a time

open a report, pick a ticker, set a customization, and then you get one number. but checking what happens when a gap-up, a green opening candle, and an IB break happen on the same day is impossible to do by hand.

the new way

an agent combines the reports for you.

tell your agent what you want to combine in plain english. the agent hits the API, combines the reports, analyzes the joint probability, attaches the sample size, and packages it as a dashboard.

how it works

paste a prompt. get a dashboard back.

AI agents have never been able to "read the market". now they can. one prompt, an API key, and your agents do all the rest... all from a single paragraph of plain English.

your prompt

using the edgeful API, pull data for ES over the last 1 year and combine the outside days and gap fillreports. on days that open above yesterday's high, how often does price reverse… then use the /dashboard skill to turn the results into a polished HTML dashboard.

↓ the agent reads it, fires the calls, and writes this

edgeful · the agent's output

dashboard the agent built. outside days × gap fill reversal chain on ES.

step 01 · what you need

an API key + an agent

an edgeful API key plus Claude, Cursor, or Claude.ai. Claude Code and Cursor work best. they iterate across many calls in one session.

step 02 · how to run it

paste it. hit enter.

the agent picks endpoints, fires the calls, does the math. part 1 is ~30s; the deep dives are a couple of minutes. every prompt ends with /dashboard.

step 03 · how to read it

every stat has a sample size

small samples (under 30) get flagged. every dashboard cites ticker, timeframe & session. no made-up numbers. every value comes from a real API call.

the kit · 10 prompts

ten prompts, organized in three parts.

every dashboard in this kit was built by an agent running these exact prompts against the edgeful API. zero numbers typed by hand.

part 1

combining 2+ reports

stack reports to find confluence. when A and B both happen, what's the probability of C?

prompt01gap + outside day reversal (both directions)joins outside-days and gap-fill data. first question: did the outside day reverse? second question: of those, how many ran all the way to a full gap fill?

prompt02opening candle + IB break + IB by closestacks three reports. does opening candle direction lift the IB break rate above baseline?

prompt03inside bars + IB break directionisolates inside-bar days and slices them by IB break behavior. how often the natural target (yesterday's high or low) gets tagged when the IB breaks the same way.

prompt04gap fill + ICT opening retracementa two-target question: when the open sits above both the prior close and the midnight open, does price retrace down far enough to tag both levels?

prompt05ORB break + IB break alignmenthow often the ORB and IB break the same direction, and when they align, whether the session continues or reverses back inside the IB.

part 2

deep dives

go wide on one report family. views the edgeful UI can't show in a single screen.

prompt06IB report deep dive: comprehensive dashboardpulls all 11 IB endpoint variants. the agent assembles a multi-section dashboard with the top edges ranked at the end.

prompt07IB confluence sweep: what actually shifts the IBrecomputes the IB break distribution under 8 filter conditions side by side, then flags any signal that shifts the baseline 5+ points.

prompt08IB by rejection: multi-dimensional breakdownthe rejection view the edgeful UI only shows one slice at a time. broken out three ways: by ending zone, by weekday, and by IB size.

part 3

setup discovery + screening

paste in the morning, get a bias for the session. one clean question: where's the edge today.

prompt09cross-market high-probability scanscans 4 reports across 4 tickers (16 API calls) and surfaces the strongest 70%+ setups, ranked, with sample sizes attached.

prompt10weekday-specific stacked biascombines three by-weekday reports into a single bias read. paste it in the morning, get the session's edge in one paragraph.

the 10 prompts are one email away.

put in your email and we'll send the kit for free. no spam, instant access.

built by edgeful · for traders who use AI · unsubscribe anytime

results in this kit are historical, sample-size-attached, and NY session only. they are not predictive. these prompts give you a fast way to ask precise questions of real market data. interpreting the answers and acting on them is on you. futures trading involves substantial risk. past performance is not indicative of future results.