AI
AI Toolkit
Use an AI coding assistant to build on the Partner API, or connect an agent to a COUNT workspace over MCP.
Choose a path
Build an integration
You are writing code that calls the Partner API from your own product. Generate a prompt for Cursor or Claude Code that covers OAuth, HMAC signing and the endpoints you need.
Automate from a coding agent
Claude Code, Cursor or another agent on your machine works in a COUNT workspace. The COUNT CLI signs you in and runs a local MCP server.
Connect a hosted assistant
Claude.ai, ChatGPT or another client that supports remote MCP. Sign in with OAuth and the assistant gets 319 workspace tools — nothing to install.
The paths combine: many teams build their integration with generated prompts, then use MCP for their own back-office work. Local and remote MCP expose the same COUNT_* tools.
Give your assistant the docs
If your AI tool accepts a documentation URL, give it one of these. Both are rebuilt from this site on every deploy.
/llms.txtAn index of every page, one line each. Small enough to give an assistant up front so it knows what exists./llms-full.txtThe whole documentation set as one markdown file. One fetch instead of crawling — best for large context windows.Grounding beats guessing
COUNT's HMAC signing scheme is specific to COUNT, and a model without the docs in context will confidently invent a plausible-looking one. Giving itllms.txt first is the cheapest fix for wrong generated code.Connect a remote MCP client
Add COUNT to an MCP client to let an assistant read and write real workspace data. The full tool catalog is on the MCP Server page.
VS Code
.vscode/mcp.json in your workspace:{
"servers": {
"count": {
"type": "http",
"url": "https://api.getcount.com/mcp"
}
}
}Cursor
~/.cursor/mcp.json:{
"mcpServers": {
"count": {
"url": "https://api.getcount.com/mcp"
}
}
}Claude Code
claude mcp add --transport http count https://api.getcount.com/mcp/mcp.Claude.ai
Settings → Connectors → Add custom connector, then paste the server URL.
ChatGPT
Open Plugins → Browse plugins, select the + button at the top, then choose Create app → MCP app. Paste the server URL as the MCP endpoint.
Any other MCP client
https://api.getcount.com/mcpCOUNT_* tools under the name count.Run MCP locally
For an agent on your machine, install the CLI, sign in, and print the MCP configuration to paste into your agent. Details on the COUNT CLI page.
npm install -g @countfinancial/cli
count login
count mcp print-config