you are setting up the edgeful API for me on this machine. edgeful is a market-data
platform for futures/stock traders. your job is to take me from zero to a working,
verified API connection, end to end. work autonomously.
IMPORTANT: never ask me to paste my API key into this chat. i'll put it directly
into the .env file myself in step 2. pause and wait for me to confirm i've done that.
================================================================
step 1 - read the docs (do this first, don't guess)
================================================================
fetch and read these pages so you know the real endpoints, parameters, and response
shapes. do not invent parameter names. confirm everything against the docs:
- welcome / overview: https://www.edgeful.com/docs/welcome
- quickstart (5-min setup): https://www.edgeful.com/docs/quickstart
- full endpoint index: https://www.edgeful.com/docs/llms.txt
- authentication + limits: https://www.edgeful.com/docs/authentication.md
- session presets: https://www.edgeful.com/docs/session-presets.md
- machine-readable spec: https://www.edgeful.com/docs/openapi-public.json
key facts you'll confirm there:
- base URL: https://api.edgeful.com
- auth header: Authorization: Bearer <my key>
- rate limits: use the current limits from the authentication docs and handle
429s with backoff
================================================================
step 2 - set up the .env file (i paste the key, not you)
================================================================
i get my API key from the edgeful API dashboard:
https://www.edgeful.com/api-dashboard
that page is where i generate my key. the key is shown only once and starts with
"ef_live_". the dashboard also has explainer videos and links to the API docs.
do not ask me to type or paste my key into this chat. keys must never go into the
conversation. instead, prepare the file so i can paste it in myself:
- check if a .env file already exists in this project:
- if it does not exist, create one with this exact placeholder line:
EDGEFUL_API_KEY=
- if it does exist, check for an EDGEFUL_API_KEY line. if missing, append the
empty placeholder line above. do not overwrite or remove any other variables.
- make sure .env is listed in .gitignore. add it if it is missing.
- then tell me: "open .env and paste your key after EDGEFUL_API_KEY=, save it, and
reply 'done'." then stop and wait for me.
- after i reply "done", load it for this session:
set -a && source .env && set +a
verify the key is present without printing it before continuing.
================================================================
step 3 - install the edgeful dashboard skill
================================================================
download and install the official dashboard skill. it turns report data into polished
single-file HTML dashboards:
https://app.edgeful.com/storage/v1/object/public/skills/dashboard-skill.zip
then adapt the install to whichever agent you are:
- if you are Claude Code:
unzip it, then copy the inner dashboard/ folder into this project's skills dir:
cp -R dashboard-skill/dashboard ./.claude/skills/
or use ~/.claude/skills/ for global use. it loads next session and is invokable
with /dashboard or "build me a dashboard from this data".
- if you are Codex, Cursor, or another coding agent:
unzip it into the project, for example ./skills/dashboard/. read its SKILL.md
and references/ as your styling and layout guide whenever i ask for a dashboard.
treat SKILL.md as the source of truth for the dark-mode design system.
================================================================
step 4 - make the first live call (prove it works)
================================================================
run an RTY previous-day's-range report for the New York session over roughly the
last 3 months. compute the dates at runtime so the setup stays current:
START_DATE=$(python3 - <<'PY'
from datetime import date, timedelta
print(date.today() - timedelta(days=92))
PY
)
END_DATE=$(python3 - <<'PY'
from datetime import date, timedelta
print(date.today() - timedelta(days=1))
PY
)
curl -s -H "Authorization: Bearer $EDGEFUL_API_KEY" \
"https://api.edgeful.com/report_calculation/previous-days-range-standard/futures/RTY?start_date=${START_DATE}&end_date=${END_DATE}&start_time=09:30:00&end_time=16:00:00&timezone=America/New_York"
what to check in the response:
- HTTP 200 and a JSON body with a summary object = success
- detailed or detailedTable rows may be absent unless my plan includes row-level detail
- 401 = key missing or revoked; re-check step 2
- 403 = plan access issue; read the code and explain which plan/ticker/report/range
limit was hit
- 429 = rate limited; wait and retry with backoff
- empty {} = check ticker, date range, market type, and session parameters
================================================================
step 5 - report back
================================================================
when done, give me a short summary:
- confirm the key is stored in .env and gitignored. do not print the key itself.
- confirm the dashboard skill is installed and how to invoke it.
- paste the headline numbers from the RTY response's summary object:
- summary.prevDayHigh count and percentage
- summary.prevDayLow count and percentage
- if present, summary.prevDayHighGreen/Red and summary.prevDayLowGreen/Red
also include the returned startDate and endDate. do not invent inside-range counts,
union break counts, totalDays, or sample size if the response does not include them.
- tell me one example command i can run next to pull a different report, such as a
gap fill report on NQ, referencing the endpoint index at
https://www.edgeful.com/docs/llms.txt.
================================================================
step 6 - offer to build a dashboard
================================================================
finally, ask me: "want me to build a simple dashboard for this RTY data using the
dashboard skill?" then stop and wait for my answer.
- if i say yes, use the dashboard skill installed in step 3 to turn the RTY
previous-day's-range response into a single-file HTML dashboard: KPI cards for
prevDayHigh and prevDayLow counts/percentages, optional close green/red breakdowns,
and the returned date range. follow the skill's dark-mode design system. save it
and tell me the file path.
- if i say no, stop here. setup is complete.
