fix(errors): guard HTML rate limit copy

altaywtf · altaywtf · commit 755cff833c1f · 2026-03-26T15:10:34.000+03:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -42,6 +42,7 @@ Docs: https://docs.openclaw.ai
 - WhatsApp: fix infinite echo loop in self-chat DM mode where the bot's own outbound replies were re-processed as new inbound user messages. (#54570) Thanks @joelnishanth
 - Discord/reconnect: drain stale gateway sockets, clear cached resume state before forced fresh reconnects, and fail closed when old sockets refuse to die so Discord recovery stops looping on poisoned resume state. (#54697) Thanks @ngutman.
 - BlueBubbles/groups: optionally enrich unnamed participant lists with local macOS Contacts names after group gating passes, so group member context can show names instead of only raw phone numbers.
+- Agents/errors: surface provider quota/reset details when available, but keep HTML/Cloudflare rate-limit pages on the generic fallback so raw error pages are not shown to users. (#54512) Thanks @bugkill3r.
 
 ## 2026.3.24
 
diff --git a/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts b/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts
@@ -155,6 +155,24 @@ describe("formatAssistantErrorText", () => {
     expect(result).toBe("⚠️ Your quota has been exhausted, try again in 24 hours");
   });
 
+  it("falls back to generic copy for HTML quota pages", () => {
+    const msg = makeAssistantError(
+      "429 <!DOCTYPE html><html><body>Your quota is exhausted</body></html>",
+    );
+    expect(formatAssistantErrorText(msg)).toBe(
+      "⚠️ API rate limit reached. Please try again later.",
+    );
+  });
+
+  it("falls back to generic copy for prefixed HTML rate-limit pages", () => {
+    const msg = makeAssistantError(
+      "Error: 521 <!DOCTYPE html><html><body>rate limit</body></html>",
+    );
+    expect(formatAssistantErrorText(msg)).toBe(
+      "⚠️ API rate limit reached. Please try again later.",
+    );
+  });
+
   it("returns a friendly message for empty stream chunk errors", () => {
     const msg = makeAssistantError("request ended without sending any chunks");
     expect(formatAssistantErrorText(msg)).toBe("LLM request timed out.");
diff --git a/src/agents/pi-embedded-helpers/errors.ts b/src/agents/pi-embedded-helpers/errors.ts
@@ -66,19 +66,30 @@ const RATE_LIMIT_SPECIFIC_HINT_RE =
   /\bmin(ute)?s?\b|\bhours?\b|\bseconds?\b|\btry again in\b|\breset\b|\bplan\b|\bquota\b/i;
 
 function extractProviderRateLimitMessage(raw: string): string | undefined {
+  const withoutPrefix = raw.replace(ERROR_PREFIX_RE, "").trim();
   // Try to pull a human-readable message out of a JSON error payload first.
-  const info = parseApiErrorInfo(raw);
+  const info = parseApiErrorInfo(raw) ?? parseApiErrorInfo(withoutPrefix);
   // When the raw string is not a JSON payload, strip any leading HTTP status
   // code (e.g. "429 ") so the surfaced message stays clean.
-  const candidate = info?.message ?? (extractLeadingHttpStatus(raw.trim())?.rest || raw);
+  const candidate =
+    info?.message ?? (extractLeadingHttpStatus(withoutPrefix)?.rest || withoutPrefix);
 
   if (!candidate || !RATE_LIMIT_SPECIFIC_HINT_RE.test(candidate)) {
     return undefined;
   }
 
+  // Skip HTML/Cloudflare error pages even if the body mentions quota/plan text.
+  if (isCloudflareOrHtmlErrorPage(withoutPrefix)) {
+    return undefined;
+  }
+
   // Avoid surfacing very long or clearly non-human-readable blobs.
   const trimmed = candidate.trim();
-  if (trimmed.length > 300 || trimmed.startsWith("{")) {
+  if (
+    trimmed.length > 300 ||
+    trimmed.startsWith("{") ||
+    /^(?:<!doctype\s+html\b|<html\b)/i.test(trimmed)
+  ) {
     return undefined;
   }
 
