TL;DR
Anthropic is a AI & ML MCP server that lets Claude Code, Cursor, Windsurf and any MCP-compatible AI agent call Anthropic Claude models, manage messages and batches from any MCP-compatible agent. Install in 1 minute with mcpizy install anthropic.
Anthropic
3 toolsAI & ML
Call Anthropic Claude models, manage messages and batches from any MCP-compatible agent.
Official homepageInstall Anthropic
mcpizy install anthropicWe could not verify a published package or a reachable endpoint for this server, so we are not showing an install command that would fail. Checked 2026-08-11.
Tools exposed
3 tools availablesend_messageCall Claude with a prompt and return the response
send_messageCall Claude with a prompt and return the response
Inputs
modelstringrequiredmessagesobject[]required
list_modelsList available Claude models
list_modelsList available Claude models
count_tokensCount tokens in a prompt for a model
count_tokensCount tokens in a prompt for a model
Inputs
modelstringrequiredmessagesobject[]required
Configuration
Works identically across clients. Only the config file path differs.
Claude Code~/.claude.json
~/.claude.json{
"mcpServers": {
"anthropic": {
"command": "npx",
"args": [
"-y",
"anthropic-mcp-server"
],
"env": {
"ANTHROPIC_API_KEY": "sk-ant-..."
}
}
}
}Cursor.cursor/mcp.json
.cursor/mcp.json{
"mcpServers": {
"anthropic": {
"command": "npx",
"args": [
"-y",
"anthropic-mcp-server"
],
"env": {
"ANTHROPIC_API_KEY": "sk-ant-..."
}
}
}
}Windsurf~/.codeium/windsurf/mcp_config.json
~/.codeium/windsurf/mcp_config.json{
"mcpServers": {
"anthropic": {
"command": "npx",
"args": [
"-y",
"anthropic-mcp-server"
],
"env": {
"ANTHROPIC_API_KEY": "sk-ant-..."
}
}
}
}Authentication setup
- 1
Create an API key in the Anthropic Console
Get itANTHROPIC_API_KEY
What you can do with Anthropic MCP
Paste any of these prompts into Claude Code, Cursor or another MCP-compatible client.
“Ask Claude Opus to draft a product launch email in 3 variations”
Uses: send_message
“Count tokens in this 5-page contract before I send it”
Uses: count_tokens
“List the Claude models I can call”
Uses: list_models
Works with
More AI & ML MCPs
Stability AI
Image generation with Stable Diffusion
Chronulus AI
Predict anything with AI forecasting. Time series predictions, anomaly detection, and trend analysis.
AgentQL
Enable AI agents to get structured data from unstructured web pages using natural language queries.
Hugging Face
Models, datasets, and spaces on HF
Alternatives to Anthropic
If Anthropic doesn't fit your stack, these AI & ML MCP servers solve similar problems.
Frequently asked questions
What is the Anthropic MCP server?
Anthropic is listed as an AI & ML Model Context Protocol server whose documented purpose is to let an MCP-compatible agent call Anthropic Claude models, manage messages and batches from any MCP-compatible agent. Client support, permissions and authentication remain specific to the upstream project.
How do I install Anthropic MCP with Claude Code?
The fastest way is the MCPizy CLI: run `mcpizy install anthropic` and MCPizy will add the server to your `.claude.json` automatically. You can also install it manually by adding an entry under `mcpServers` in `.claude.json` with the command `` and restarting Claude Code.
Does MCPizy charge to view or install Anthropic MCP?
No. Reading this profile and copying its install guidance is free. Upstream software, API, account or usage costs are separate, and licence coverage should be checked from the linked source when available.
Does Anthropic MCP work with Cursor and Windsurf?
The MCP standard is supported by those clients, but a specific server can still have client-specific setup or transport limits. Use the configuration shown on this page when present and verify the upstream instructions before installation.
What can I do with Anthropic MCP?
Its documented purpose is to let an MCP-compatible agent call Anthropic Claude models, manage messages and batches from any MCP-compatible agent. The exact tools, permissions and write capabilities depend on the server version and configuration shown by the upstream project.
Compare before you commit
Put this server beside an alternative, or return to the full evidence directory.
Compare MCP serversBack to directory