ruvnet/ruflo · error

Failed to fetch base servers

Error message

Failed to fetch base servers: ${response.statusText}

What it means

Thrown inside refreshMcpServers(), the client-side Svelte store that merges MCP servers from the app's own API (${base}/api/mcp/servers) with locally-stored custom servers and the WASM server. A non-OK HTTP status aborts the refresh; the surrounding try/catch records the error in store state, so the user sees a stale or empty server list.

Solutions

1. Reproduce in the browser: open ${base}/api/mcp/servers directly and read the status/body.
2. 401/403: re-login or refresh the session; verify MCP_FORWARD_HF_USER_TOKEN and the user's HF token if server-side overlay auth is enabled.
3. 500: check server logs for the MCP hub error (often the upstream MCP host or MCP_SERVERS env parse).
4. 404/misroute: fix the reverse proxy so /api/mcp/servers reaches the SvelteKit server unchanged.
5. Reload the page after fixing — refreshMcpServers is called on init and will repopulate the store.

// probe before relying on the store
const res = await fetch(`${base}/api/mcp/servers`);
if (!res.ok) {
	throw new Error(`MCP servers endpoint returned ${res.status}`);
}

try {
	await refreshMcpServers();
} catch (err) {
	// store keeps previous allMcpServers value; surface error state to the UI
	console.error("mcp refresh failed", err);
}

When it happens

Trigger: Browser fetch to /api/mcp/servers returns non-2xx: 401/403 when the session expired or MCP_FORWARD_HF_USER_TOKEN forwarding rejected the token; 500 when server-side MCP hub configuration (MCP_SERVERS) is broken; 404 after a route change or bad `base` in a reverse-proxy deployment.

Common situations: Logged-out or expired-cookie tab refreshing MCP servers; reverse proxy stripping or misrouting /api/mcp/*; server-side MCP hub (Hugging Face MCP host) outage; deploying front and back on different origins without forwarding the path.

Understand the failure class

Background: 'Something went wrong' / 'Request failed (500)' / 'HTTP error! status: 404' — what failed HTTP requests actually mean and how to find the real cause — this error's family across 28 libraries.

Related errors

• FORBIDDEN_HEADER
• FORBIDDEN_PROTOCOL
• HTTP transport already running
• HTTP transport requires host and port configuration
• INVALID_HEADER_VALUE

AI-assisted analysis of ruvnet/ruflo@fa13ee4ad6 (2026-08-18). Data as JSON: /api/errors/f148691bf8d189af. Report an issue: GitHub.

Appendix: source

Thrown at ruflo/src/ruvocal/src/lib/stores/mcpServers.ts:171

export const allBaseServersEnabled = derived(
	[allMcpServers, selectedServerIds],
	([$all, $selected]) => {
		const baseServers = $all.filter((s) => s.type === "base");
		return baseServers.length > 0 && baseServers.every((s) => $selected.has(s.id));
	}
);

// Note: Authorization overlay (with user's HF token) for the Hugging Face MCP host
// is applied server-side when enabled via MCP_FORWARD_HF_USER_TOKEN.

/**
 * Refresh base servers from API and merge with custom servers + WASM server
 */
export async function refreshMcpServers() {
	try {
		const response = await fetch(`${base}/api/mcp/servers`);
		if (!response.ok) {
			throw new Error(`Failed to fetch base servers: ${response.statusText}`);
		}

		const baseServers: MCPServer[] = await response.json();
		const customServers = loadCustomServers();

		// Create WASM server and add to the list
		const wasmServer = createWasmServer();

		// Merge base, custom, and WASM servers
		const merged = [wasmServer, ...baseServers, ...customServers];
		allMcpServers.set(merged);

		// Load disabled base servers
		const disabledBaseIds = loadDisabledBaseIds();

		// Auto-enable all base servers that aren't explicitly disabled
		// Plus keep any custom servers that were previously selected
		// WASM server is auto-enabled by default

View on GitHub (pinned to fa13ee4ad6)