fix(onboarding): wait for gateway before health

src/commands/configure.wizard.ts

@@ -9,7 +9,7 @@ import { ensureControlUiAssetsBuilt } from "../infra/control-ui-assets.js";
 import type { RuntimeEnv } from "../runtime.js";
 import { defaultRuntime } from "../runtime.js";
 import { note } from "../terminal/note.js";
-import { resolveUserPath, sleep } from "../utils.js";
+import { resolveUserPath } from "../utils.js";
 import { createClackPrompter } from "../wizard/clack-prompter.js";
 import { WizardCancelledError } from "../wizard/prompts.js";
 import { removeChannelConfigWizard } from "./configure.channels.js";
@@ -41,6 +41,7 @@ import {
   probeGatewayReachable,
   resolveControlUiLinks,
   summarizeExistingConfig,
+  waitForGatewayReachable,
 } from "./onboard-helpers.js";
 import { promptRemoteGatewayConfig } from "./onboard-remote.js";
 import { setupSkills } from "./onboard-skills.js";
@@ -354,7 +355,22 @@ export async function runConfigureWizard(
       }
 
       if (selected.includes("health")) {
-        await sleep(1000);
+        const localLinks = resolveControlUiLinks({
+          bind: nextConfig.gateway?.bind ?? "loopback",
+          port: gatewayPort,
+          customBindHost: nextConfig.gateway?.customBindHost,
+          basePath: undefined,
+        });
+        const remoteUrl = nextConfig.gateway?.remote?.url?.trim();
+        const wsUrl = nextConfig.gateway?.mode === "remote" && remoteUrl ? remoteUrl : localLinks.wsUrl;
+        const token = nextConfig.gateway?.auth?.token ?? process.env.CLAWDBOT_GATEWAY_TOKEN;
+        const password = nextConfig.gateway?.auth?.password ?? process.env.CLAWDBOT_GATEWAY_PASSWORD;
+        await waitForGatewayReachable({
+          url: wsUrl,
+          token,
+          password,
+          deadlineMs: 15_000,
+        });
         try {
           await healthCommand({ json: false, timeoutMs: 10_000 }, runtime);
         } catch (err) {
@@ -459,7 +475,24 @@ export async function runConfigureWizard(
         }
 
         if (choice === "health") {
-          await sleep(1000);
+          const localLinks = resolveControlUiLinks({
+            bind: nextConfig.gateway?.bind ?? "loopback",
+            port: gatewayPort,
+            customBindHost: nextConfig.gateway?.customBindHost,
+            basePath: undefined,
+          });
+          const remoteUrl = nextConfig.gateway?.remote?.url?.trim();
+          const wsUrl =
+            nextConfig.gateway?.mode === "remote" && remoteUrl ? remoteUrl : localLinks.wsUrl;
+          const token = nextConfig.gateway?.auth?.token ?? process.env.CLAWDBOT_GATEWAY_TOKEN;
+          const password =
+            nextConfig.gateway?.auth?.password ?? process.env.CLAWDBOT_GATEWAY_PASSWORD;
+          await waitForGatewayReachable({
+            url: wsUrl,
+            token,
+            password,
+            deadlineMs: 15_000,
+          });
           try {
             await healthCommand({ json: false, timeoutMs: 10_000 }, runtime);
           } catch (err) {

src/commands/onboard-helpers.ts

@@ -17,7 +17,7 @@ import { runCommandWithTimeout } from "../process/exec.js";
 import type { RuntimeEnv } from "../runtime.js";
 import { stylePromptTitle } from "../terminal/prompt-style.js";
 import { GATEWAY_CLIENT_MODES, GATEWAY_CLIENT_NAMES } from "../utils/message-channel.js";
-import { CONFIG_DIR, resolveUserPath } from "../utils.js";
+import { CONFIG_DIR, resolveUserPath, sleep } from "../utils.js";
 import { VERSION } from "../version.js";
 import type { NodeManagerChoice, OnboardMode, ResetScope } from "./onboard-types.js";
 
@@ -333,6 +333,38 @@ export async function probeGatewayReachable(params: {
   }
 }
 
+export async function waitForGatewayReachable(params: {
+  url: string;
+  token?: string;
+  password?: string;
+  /** Total time to wait before giving up. */
+  deadlineMs?: number;
+  /** Per-probe timeout (each probe makes a full gateway health request). */
+  probeTimeoutMs?: number;
+  /** Delay between probes. */
+  pollMs?: number;
+}): Promise<{ ok: boolean; detail?: string }> {
+  const deadlineMs = params.deadlineMs ?? 15_000;
+  const pollMs = params.pollMs ?? 400;
+  const probeTimeoutMs = params.probeTimeoutMs ?? 1500;
+  const startedAt = Date.now();
+  let lastDetail: string | undefined;
+
+  while (Date.now() - startedAt < deadlineMs) {
+    const probe = await probeGatewayReachable({
+      url: params.url,
+      token: params.token,
+      password: params.password,
+      timeoutMs: probeTimeoutMs,
+    });
+    if (probe.ok) return probe;
+    lastDetail = probe.detail;
+    await sleep(pollMs);
+  }
+
+  return { ok: false, detail: lastDetail };
+}
+
 function summarizeError(err: unknown): string {
   let raw = "unknown error";
   if (err instanceof Error) {

src/commands/onboard-non-interactive/local.ts

@@ -1,13 +1,14 @@
 import type { ClawdbotConfig } from "../../config/config.js";
 import { CONFIG_PATH_CLAWDBOT, resolveGatewayPort, writeConfigFile } from "../../config/config.js";
 import type { RuntimeEnv } from "../../runtime.js";
-import { sleep } from "../../utils.js";
 import { DEFAULT_GATEWAY_DAEMON_RUNTIME } from "../daemon-runtime.js";
 import { healthCommand } from "../health.js";
 import {
   applyWizardMetadata,
   DEFAULT_WORKSPACE,
   ensureWorkspaceAndSessions,
+  resolveControlUiLinks,
+  waitForGatewayReachable,
 } from "../onboard-helpers.js";
 import type { OnboardOptions } from "../onboard-types.js";
 
@@ -88,8 +89,17 @@ export async function runNonInteractiveOnboardingLocal(params: {
 
   const daemonRuntimeRaw = opts.daemonRuntime ?? DEFAULT_GATEWAY_DAEMON_RUNTIME;
   if (!opts.skipHealth) {
-    await sleep(1000);
+    const links = resolveControlUiLinks({
-    // Health check runs against the gateway; small delay avoids flakiness during install/start.
+      bind: gatewayResult.bind as "auto" | "lan" | "loopback" | "custom",
+      port: gatewayResult.port,
+      customBindHost: nextConfig.gateway?.customBindHost,
+      basePath: undefined,
+    });
+    await waitForGatewayReachable({
+      url: links.wsUrl,
+      token: gatewayResult.gatewayToken,
+      deadlineMs: 15_000,
+    });
     await healthCommand({ json: false, timeoutMs: 10_000 }, runtime);
   }
 

src/wizard/onboarding.finalize.ts

@@ -14,6 +14,7 @@ import {
   formatControlUiSshHint,
   openUrl,
   probeGatewayReachable,
+  waitForGatewayReachable,
   resolveControlUiLinks,
 } from "../commands/onboard-helpers.js";
 import type { OnboardOptions } from "../commands/onboard-types.js";
@@ -31,7 +32,7 @@ import { isSystemdUserServiceAvailable } from "../daemon/systemd.js";
 import { ensureControlUiAssetsBuilt } from "../infra/control-ui-assets.js";
 import type { RuntimeEnv } from "../runtime.js";
 import { runTui } from "../tui/tui.js";
-import { resolveUserPath, sleep } from "../utils.js";
+import { resolveUserPath } from "../utils.js";
 import type { GatewayWizardSettings, WizardFlow } from "./onboarding.types.js";
 import type { WizardPrompter } from "./prompts.js";
 
@@ -209,7 +210,18 @@ export async function finalizeOnboardingWizard(options: FinalizeOnboardingOption
   }
 
   if (!opts.skipHealth) {
-    await sleep(1500);
+    const probeLinks = resolveControlUiLinks({
+      bind: nextConfig.gateway?.bind ?? "loopback",
+      port: settings.port,
+      customBindHost: nextConfig.gateway?.customBindHost,
+      basePath: undefined,
+    });
+    // Daemon install/restart can briefly flap the WS; wait a bit so health check doesn't false-fail.
+    await waitForGatewayReachable({
+      url: probeLinks.wsUrl,
+      token: settings.gatewayToken,
+      deadlineMs: 15_000,
+    });
     try {
       await healthCommand({ json: false, timeoutMs: 10_000 }, runtime);
     } catch (err) {