Last verified: 2026-08-05
The xrpldashboard MCP server is live at
https://mcp.xrpldashboard.com/mcp.
Fifteen read-only tools over XRPL, every response wrapped in a proof-annotation envelope pointing back to the source of truth. Free, no account, no key. Public beta through 2026-09.
Public beta through 2026-09.
Running on our own rippled full-history node from a Lenovo box in Indiana, exposed via a Cloudflare Tunnel. Session-scoped rate limit
600 tool calls / hour / session, enforced live — over-cap requests get HTTP 429 with a
Retry-After hint. No throttle-then-degrade, no silent drops.
Claude Desktop supports remote MCP servers two ways. Either works — pick the one that matches your setup.
xrpldashboardhttps://mcp.xrpldashboard.com/mcpxrpldashboard connector on your next new conversation.
If your MCP client edits claude_desktop_config.json
directly, or you're on a client that doesn't yet support remote MCP natively, the
mcp-remote
bridge works today:
{
"mcpServers": {
"xrpldashboard": {
"command": "npx",
"args": [
"mcp-remote@latest",
"https://mcp.xrpldashboard.com/mcp"
]
}
}
}
Restart your MCP client after saving. First run of npx mcp-remote
downloads the bridge; subsequent runs are cached.
Three prompts that exercise the shape of the server. One primitive read, one aggregation, and one moat demo — a signed daily snapshot you can verify without trusting us.
get_ledger_stats.
Live rippled result, freshness ≤ 5min. Source = local_rippled in the proof envelope.
get_rlusd_supply and get_rlusd_flow_24h.
Second is finalized-window only — never a partial-day number reported as a full one. The freshness contract is on the envelope, not buried in a footnote.
get_signed_snapshot then verify_snapshot_signature.
The tool re-verifies against the pinned pubkey at
/.well-known/snapshots/pubkey.pem. If it validates, you now know the day's snapshot came from our key, not from something between you and us.
/openapi.json.
proof.source,
proof.as_of, a freshness_contract, and a
claims_ref you can dereference at
/claims. When we can't promise strict, we return
honest_partial with the reason.
Retry-After
seconds, or start a new client session (which starts a fresh bucket).
/health and
/walker_health. Every rate-limit block writes a walker_health row so misuse patterns show up on the fleet health page, not silently on our side.