How to use the Liquidation Heatmap Analysis agent skill

What is an agent skill and what does this one make the agent do?

An agent skill is a folder containing a SKILL.md file with instructions and, optionally, code that an AI agent can load when a task matches the skill's description. This skill's file is named SKILL.md and is described as liquidation level analysis and heatmap interpretation.

When the skill is active, the agent can identify where leveraged position concentration sits, discuss liquidation cascades and stop-hunt zones, and treat liquidation data as support or resistance signals. The file also explains liquidation mechanics: for a long position, the liquidation price is estimated as entry price times (1 minus 1 divided by leverage plus maintenance margin), and the mirror formula applies to shorts.

• Explains how liquidation prices are calculated from entry price, leverage and maintenance margin.
• Describes how to read a heatmap where closures are concentrated by price level.
• Frames clusters as magnets, cascade triggers, or later support and resistance.
• Includes sample Python functions for estimating clusters and generating simple magnet, cascade risk and post-liquidation signals.

How do you install the liquidation-heatmap skill?

Download the ZIP file from this page. Unzip it into your personal skills folder at ~/.claude/skills/ so the skill is available across projects, or into a project's .claude/skills/ folder to limit it to that project.

If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: Settings → Capabilities → Skills, then upload

How do you trigger the skill with natural requests?

You do not need a special command. Ask about liquidation levels, heatmaps or leveraged positions in plain language, and the agent can match your request to the skill's description. Naming the concepts in the skill, such as liquidation clusters, cascade risk or stop-hunt zones, makes a match more likely.

Give the agent the data it needs. The skill does not supply market data, so paste or describe the open interest and leverage figures you have, along with the exchange and price levels you care about.

• "Explain how the liquidation price for a 10x BTC long at $65,000 is calculated."
• "Read this liquidation heatmap and tell me where the largest long clusters sit."
• "Is there cascade risk below the current price given these cluster levels?"
• "Which liquidation clusters could act as support or resistance after a big wipeout?"

What examples does the skill contain?

The SKILL.md file works through a worked example: a BTC long opened at $65,000 with 10x leverage and a 0.5% maintenance margin has an estimated liquidation price of $58,825, meaning a 9.5% move against the position triggers liquidation. It also includes a leverage table showing that liquidation distance shrinks as leverage rises, from about 50% at 2x to about 1% at 100x.

It includes a sample heatmap reading table with price levels labelled as heavy or moderate long and short liquidation zones, and a list of metrics such as 24h total liquidations and the long/short liquidation ratio, with interpretation bands.

What do you need before using it?

You need Python to run the example functions. You also need liquidation and open-interest data from an exchange or data provider that you supply yourself. The skill does not name a specific data source or API, and the functions it contains are placeholders that depend on the data you provide.

The skill does not run scripts and does not use the network, so it only gives the agent written instructions and example code to reason with.

• Python, to run the example functions
• Your own liquidation and open-interest data from an exchange or data provider
• Licence: MIT

What are the safety notes?

The file presents liquidation magnets and stop-hunt zones as fact, but it gives no backtest or evidence that these signals predict price. Treat the outputs as analysis prompts, not forecasts.

The example functions are placeholders. The clusters they produce depend entirely on the inputs you feed them, and the thresholds in the code, such as the $100M liquidation cut-off and the 2% and 5% gap rules, are examples rather than validated values.

• Do not make real trading decisions with money you cannot afford to lose based on this skill alone.
• Check any signal against your own data before relying on it.
• The skill runs no scripts and uses no network access, so it cannot fetch live liquidation data itself.

What tips help you get better results?

Ask the agent to state its assumptions, such as the leverage distribution and maintenance margin it used. Because the cluster functions depend on estimated inputs, results are only as good as the data behind them.

Use the skill to learn mechanics. The liquidation formula and leverage table are a clear way to see how leverage and maintenance margin affect where a position is closed.

• Always supply the leverage and maintenance margin values you are assuming.
• Ask for the formula so you can check the arithmetic yourself.
• Compare the agent's cluster reading against an actual heatmap from your data source.