MCP Server
Axiora speaks MCP, so an AI agent can query it directly, 41 read-only tools over the same EDINET-sourced data as the REST API. Point a client at one URL, use your existing key, and the model calls the tools itself.
What it is
Every tool is read-only: an agent reads data, never writes or changes it. They run on the same backend and the same key as the REST API, so what an agent reports back is exactly what the API returns.
Use MCP when you want a model to do the querying; use REST when your own code does. Same data either way.
Endpoint & transport
Point any MCP-compatible client at the endpoint below. The transport is Streamable HTTP.
Authentication
Authentication uses the same Bearer token as the REST API, passed in the Authorization header. The MCP server needs a Developer or Institutional key.
Authorization: Bearer ax_live_•••MCP needs no separate key. It uses the same ax_live_ token as REST. See Authentication for how to get one and what each tier unlocks.
Connect a client
With Claude Code you can add the server in one line:
claude mcp add --transport http axiora \
https://api.axiora.dev/mcp \
--header "Authorization: Bearer ax_live_YOUR_KEY"For other clients (Claude Desktop, Cursor, …) add the server to your MCP config (e.g. claude_desktop_config.json or .cursor/mcp.json):
{
"mcpServers": {
"axiora": {
"type": "url",
"url": "https://api.axiora.dev/mcp",
"headers": {
"Authorization": "Bearer ax_live_YOUR_KEY"
}
}
}
}Then just ask, “Show me Toyota's financials”, and the agent chains the tools itself: search_companies → get_financials.
Tools
All 41 tools, with descriptions generated straight from the live server, so this table can't drift from what the agent can actually call. You don't invoke these yourself; the agent picks one per question.