Anthropic
Claude Code
Anthropic's terminal coding agent. Adds DataFeeds with a single datafeeds entry in .mcp.json — your token never touches the file.
Pick your agent
Select an agent to get step-by-step setup instructions.
Cursor
Cursor
AI-first IDE built on VS Code. Connects to DataFeeds through .cursor/mcp.json in one project or globally.
OpenAI
Codex
OpenAI's coding agent. The CLI, IDE extension, and ChatGPT desktop app share one ~/.codex/config.toml, which names your token's env var — never its value.
What you're connecting
Three separate things carry the DataFeeds name, and they do different jobs.
MCP server
A live tool connection. Your agent gains a set of DataFeeds tools it can call mid-conversation and get real data back. Almost always what you want — it's what this guide sets up.
Sports AI Skill
Packaged instructions that teach an agent the API's conventions — sport codes, season semantics, per-sport endpoints — for code written against the REST API directly. One command in each agent guide enables it.
Documentation
The reference both humans and agents read, including the OpenAPI spec. For whenever you need exact request and response shapes.
The MCP server and the skill are complementary: the server gives your agent live data, the skill gives it judgment about how to ask. Each agent guide covers both.