When you go through a stretch of the footprint, it's easy to see what you want to see, especially after a losing trade.
The skill makes the agent work in order: facts with time and price first, then the interpretation, what would disprove it and what the data can't show.
How to install
Copy the skill text and paste it into your agent's chat with this line on top:
Save this as a skill named "orderflow-read". If you can't save skills, follow it for this conversation.
Skill text
---
name: orderflow-read
description: Read an ATAS chart by volume and order flow through the ATAS MCP server: numbers from the footprint, delta and profile first, then the interpretation separately. Use when the user asks to read the chart, explain what is happening on the footprint, or go through a bar or a stretch of the chart ("what's on this bar", "explain the last hour by delta", "was there absorption"). Not for signals, forecasts or orders.
compatibility: Requires the ATAS MCP server and a connection to a crypto exchange with footprint data. AI Connect gives the agent market data for crypto.
metadata:
version: 0.1
updated: 2026-11-01
---
# orderflow-read
Read the selected stretch of the chart and show what the data shows. Answer in the user's language. Always keep facts and interpretation apart. Do not say "buy" or "sell", and do not give targets or stops. Do not place orders.
## Hard rules
- Numbers first, then words. Every statement rests on a specific bar: time and price.
- Read numbers from tools. Use a screenshot only to check how the chart looks. Never read values from it.
- Do not name a pattern unless you can show its condition in numbers. Do not write "absorption" without figures.
- If data is missing, for example the footprint is unavailable for this stretch, say so plainly.
- Phrase the interpretation as a hypothesis: "this looks like", "this is consistent with". Name what would disprove it.
## 1. Define the stretch
Unless the user says otherwise, use the active chart and the last 30 closed bars. Do not include the bar that is still forming in your conclusions.
## 2. Read the data
1. `chart_get_snapshot` for the chart: bars, indicator values, drawing objects.
2. `chart_get_bars` with `format: "table"` and `fields`: time, OHLC, volume, delta. Note `dataSource`. If the data is delayed, say so in the answer.
3. Footprint: `chart_get_bars` with `footprint: true` and `footprintBars` no more than 100 per call. Bars are counted from the newest one returned, so set `to` to the end of the stretch. If volume is nearly equal across all price levels, say so and draw no conclusions from the levels inside the bar.
4. Profile: "Volume Profile & TPO" via `indicators_get_values`, if it is on the chart. If it isn't, suggest adding it, but do not add it yourself.
If ATAS does not return market data for the instrument, say that through AI Connect it is available for crypto, and stop.
## 3. What to calculate
- bar volume relative to the 20-bar average;
- bar delta and its sign relative to the candle color;
- cumulative delta over the stretch and the price move over the same stretch;
- where the largest volume is inside the bar: top, bottom, middle;
- levels with large volume and how price reacted to them.
## 4. Answer
```
Chart: <instrument, period, source, delay>
Range: <from and to, number of bars>
What the data shows:
- <time, price: fact with a number>
- <time, price: fact with a number>
How it can be read:
- <hypothesis>. Disproved by: <what would have to happen>
What the data doesn't show: <limitation of this reading>
```
Three to five facts. One or two hypotheses. No conclusion about where price will go.
End with one line: this is a description of the data; the trading decision is the trader's.
## 5. Markup on request
If the user asks to show it on the chart, add objects with `drawing_objects_add` with a label and time. Say that they can be removed with the phrase "remove the markup".
How to run it
Use orderflow-read. On my BTCUSDT 5m chart, go through the bar at 15:50 UTC on November 5 and the ten bars after it. Was there absorption of selling?
What you get
The answer follows the skill's order: facts by bar, what they might mean, what would disprove that reading and what the footprint can't show.
Use orderflow-read. On my BTCUSDT 5m chart, go through the bar at 15:50 UTC on November 5 and the ten bars after it. Was there absorption of selling?
Chart: BTCUSDT, 5m, Binance Futures. Source: Binance Only Market Data, not delayed. Range: Nov 5, 2026, 15:50-16:40 UTC, 11 closed bars; the last one closed at 16:45.
What the data shows:
- 15:50: open 76,160, low 75,530, close 75,750. Volume 9,319.088, which is 2.07 times the average volume of the previous 20 bars (4,511.808). Delta −1,032.622 matches the down candle; the close is 220 above the low.
- 15:50, bottom of the bar: the largest volume at one level is 511.051 at 75,600, of which bid 242.285 and ask 268.766. Across 75,610-75,630 the total is bid 685.262 against ask 486.196, delta −199.066. The close at 75,750 ended up above this zone.
- 15:55: low 75,650, close 76,010, delta +1,344.189. The next pullback: at 16:00 at 75,800 there was bid 248.881 against ask 16.547; at 16:05 at 75,700, bid 234.027 against ask 67.664. These bars closed above those levels, at 75,840 and 75,810.
- 15:55-16:40: none of the ten following bars reached the low of 75,530; the lowest low is 75,640 at 16:05. At 16:10 the close was 76,320, above the source bar's open, with delta +1,122.271.
- Whole range: total delta +1,967.142; from the 15:50 open to the 16:40 close price rose from 76,160 to 76,430, by 270. The ten following bars have a total delta of +2,999.764.
How it can be read:
- This looks like local absorption of aggressive selling at the bottom of the 15:50 bar, especially around 75,610-75,630: trades at the bid dominated, but the bar closed above the zone, and the next ten bars did not make a new low. The idea that this zone holds would be disproved if price went through 75,530 and stayed below it while delta remained negative.
- At the same time, the recovery came with positive delta of +2,999.764 after 15:50. This also fits active buying after the drop; the whole bounce can't be attributed to absorption alone.
What the data doesn't show: the footprint contains executed trades, but it doesn't prove that a limit buyer replenished orders or that there was an iceberg. So absorption here is a reasonable hypothesis; it is not an established fact. The Volume Profile & TPO indicator is not on the chart; to check the profile context, you can add it.
This is a description of the data; the trading decision is the trader's.
What you need
- A crypto exchange connection with footprint data
If there is a profile indicator on the chart, the agent takes it into account too. If there is no profile, the agent suggests adding one.
This is a description of the data. The trading decision is yours.