Query the public index from your LLM
ARC Report ships a remote Model Context Protocol server over Streamable HTTP. No API key, no account — per-IP rate limits only (60 requests/minute). Every response includes a source_url and last_updated timestamp for citation.
Public server URL
https://www.arcreport.ai/api/mcpTools
get_brand_status(domain)Per-agent access for all 9 tracked agents, platform, structured data quality, llms.txt, ARC Score, last scanned.
search_brands(query?, category?, platform?, blocking_agent?, allowing_agent?)Filtered, paginated search over the full index.
get_recent_changes(days=7, category?)Confirmed changelog entries with before/after values.
get_agent_stats(agent?)% of the index blocking each agent, by category, with week-over-week deltas.
compare_brands(domains[])Side-by-side matrix for up to 10 brands.
Resources: arc://methodology (scan design + score formula) and arc://insights/latest (headline stats from the latest scan).
Claude.ai (custom connector)
- Open Settings → Connectors → Add custom connector.
- Name it ARC Report and paste the server URL above.
- Leave authentication empty (none required) and save.
- Enable the connector in any chat via the search-and-tools menu.
Claude Desktop
Settings → Developer → Edit Config, then add to claude_desktop_config.json:
{
"mcpServers": {
"arc-report": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://www.arcreport.ai/api/mcp"]
}
}
}(Desktop launches local servers; mcp-remote bridges to the remote URL. On paid plans you can instead add it as a custom connector like Claude.ai above.)
Claude Code
claude mcp add --transport http arc-report https://www.arcreport.ai/api/mcpThen ask Claude Code anything about the dataset — it will call the tools directly.
Three prompts to try
Raw protocol (curl)
curl -X POST https://www.arcreport.ai/api/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_brand_status","arguments":{"domain":"nike.com"}}}'