Press Release

Crypto Exchanges Are Racing to Build AI Agent Skills. Here Is Where Things Stand.

Crypto

A new category is forming at the intersection of AI agents and crypto trading infrastructure. BitMart is among the first to open it up with a fully open-source approach.

The Race Nobody Was Talking About Six Months Ago

Something quiet has been happening across the crypto exchange landscape. One by one, major platforms have been shipping what they call Skills: modular integrations that give AI agents direct access to trading infrastructure. Binance has one. OKX has one. Bitget has one. Gate.io has one.

The trigger is the rise of agentic AI frameworks, most notably OpenClaw, which lets users run AI agents capable of taking real-world actions, not just answering questions. Once it became clear that millions of users were pointing these agents at financial tasks, exchanges moved fast. The question was no longer whether to build an AI Skills layer, but how.

BitMart, a global cryptocurrency exchange with over 13 million users, entered this race with the launch of BitMart Skills, an open marketplace that gives AI agents native access to its trading ecosystem. What sets it apart is its approach: fully open-source, framework-agnostic, and built around a skill format that any agent can read and execute without custom configuration.

What Exchange Skills Actually Do

The concept is straightforward. An AI agent on its own has no way to interact with a crypto exchange. It can discuss markets, analyze charts, and suggest strategies, but it cannot place an order or check a balance. Skills close that gap.

A Skill is essentially a structured instruction set, packaged in a format the agent can read, that tells it how to authenticate with an exchange, what operations are available, and how to execute them safely. Once installed, the agent gains the ability to act, not just advise.

In practice, this means a user can tell their AI agent “Buy 100 USDT worth of BTC on BitMart” or “Set a stop-loss on my ETH position at $2,800” and the agent handles the rest. No manual order entry, no switching between apps, no code required.

What BitMart Skills Covers

The BitMart Skills marketplace currently offers two production-ready modules:

Spot trading: market and limit orders, batch orders, order cancellation and management, account balance queries, margin trading, and fee rate lookup.

USDT perpetual futures: opening and closing positions, leverage management, take-profit and stop-loss orders, conditional (plan) orders, trailing orders, and position mode switching.

Together, these modules cover the full trading cycle. An agent can research market conditions, evaluate an entry point, execute a position, and monitor it over time without the user needing to touch the exchange interface at all.

The Open-Source Difference

Most exchange Skills implementations are proprietary. They work within a specific platform’s ecosystem and require the agent to be running in a particular environment. BitMart Skills takes the opposite approach.

The entire codebase is publicly available on GitHub under an open license. The skill format is readable by any AI agent framework, including Claude Code, Codex, OpenClaw, LangChain, and CrewAI. Developers can fork it, extend it, and build on top of it. Installation takes a single command.

This matters for users because it means BitMart Skills will work wherever their agent runs, not just in one walled garden. It also means the community can contribute new skills and improvements over time, something closed implementations cannot offer.

Security Built Into the Design

Connecting an AI agent to a live trading account raises legitimate concerns. BitMart Skills addresses these at the architecture level rather than leaving them to individual users to manage.

All asset-affecting operations require explicit user confirmation before execution. API permissions are scoped to read and trade access only, with withdrawal permissions deliberately excluded from the recommended configuration. Credentials are stored locally and never transmitted through chat interfaces.

The result is a system where the agent can act efficiently within defined boundaries, but cannot take actions the user has not directly authorized.

Early Days of a Larger Shift

The exchange Skills race is still in its early stages. The platforms that have shipped integrations are building toward a world where AI agents are first-class participants in crypto markets, able to execute complex strategies autonomously while keeping the user informed and in control.

BitMart Skills represents one of the more accessible entry points into this ecosystem. It is free to use, open to inspect, and compatible with the tools most developers and serious traders are already running. For anyone who has been watching the agentic AI space and wondering when it would get serious about crypto, that moment appears to have arrived.

GitHub: https://github.com/bitmartexchange/bitmart-skills

Learn more: https://www.bitmart.com

Comments
To Top

Pin It on Pinterest

Share This