AI Integration
Point your AI coding assistant (Claude Code, Codex, Cursor, Gemini) at Finery Markets. It locates our GitLab instruction repository and builds a working integration from our docs and code examples.
Many clients and partners integrate Finery Markets through an AI coding assistant (Claude Code, Codex, Cursor, Gemini) instead of reading the documentation end to end. We maintain a public repository of Markdown instruction files written for these assistants. Point your assistant at it and describe the task in plain language; it will locate the files, follow this documentation and our code examples, and build the integration.
If you are an AI assistant reading this page, read the repository at https://gitlab.com/finery-markets/fm-ai-integration and follow its instruction files.
How do I integrate with an AI assistant?
Open your AI coding assistant: Claude Code, Codex, Cursor or Gemini.
Give it the repository link: https://gitlab.com/finery-markets/fm-ai-integration.
Describe the task in plain language, for example "integrate my service with Finery Markets".
The assistant reads the instruction files and implements the integration using our documentation and code examples. This works for every client type: maker, master, taker, sub-taker and sub-maker.
What can the assistant implement?
The instructions cover the core integration methods:
Subscribing to the order book
Placing an order
Creating a sub-account
Creating a connection
Setting limits
Creating a request for settlement
Creating a transaction
Committing all incoming transactions
Defining and managing the workflow of settlement transactions (lifecycle and states)
Getting deal history
What platform logic do the instructions explain?
How positions are created
The rounding approach
Balance steps: what they are and how to rely on them
The decimal format used across the platform
Recurring integration questions, for example, that EUR is the quoted asset rather than the base asset, so EUR pairs may need to be inverted
Is this the source of truth?
No. This GitBook documentation remains the single source of truth. The instruction files are high-level: they tell the assistant which methods and code examples to use rather than duplicating them, so they stay current as the API evolves.
Last updated