Quiet noisy Telegram gateway errors

gateway/run.py

@@ -66,6 +66,144 @@ _PLATFORM_CONNECT_TIMEOUT_SECS_DEFAULT = 30.0
 _ADAPTER_DISCONNECT_TIMEOUT_SECS_DEFAULT = 5.0
 _TELEGRAM_COMMAND_MENTION_RE = re.compile(r"(?<![\w:/])/([A-Za-z0-9][A-Za-z0-9_-]*)")
 
+_TELEGRAM_NOISY_STATUS_RE = re.compile(
+    r"("  # transient/auxiliary status that should stay in logs, not Telegram chat
+    r"auxiliary\s+.+\s+failed"
+    r"|compression\s+summary\s+failed"
+    r"|fallback\s+context\s+marker"
+    r"|configured\s+compression\s+model\s+.+\s+failed"
+    r"|no\s+auxiliary\s+llm\s+provider\s+configured"
+    r"|auto-lowered\s+compression\s+threshold"
+    r"|preflight\s+compression"
+    r"|rate\s+limited\.\s+waiting\s+\d"
+    r"|retrying\s+in\s+\d"
+    r"|max\s+retries\s+\(\d+\).*(?:trying\s+fallback|exhausted|invalid\s+responses)"
+    r"|stream\s+(?:drop|drop\s+mid\s+tool-call).+retry\s+\d"
+    r"|stale\s+connections\s+from\s+a\s+previous\s+provider\s+issue"
+    r")",
+    re.IGNORECASE | re.DOTALL,
+)
+
+_GATEWAY_PROVIDER_ERROR_RE = re.compile(
+    r"("  # infrastructure/provider error preambles, not ordinary assistant prose
+    r"api\s+(?:call\s+)?failed"
+    r"|provider\s+authentication\s+failed"
+    r"|non-retryable\s+error"
+    r"|rate\s+limited\s+after\s+\d+\s+retries"
+    r"|error\s+code\s*:"
+    r"|\bhttp\s*\d{3}\b"
+    r"|incorrect\s+api\s+key"
+    r"|invalid\s+api\s+key"
+    r")",
+    re.IGNORECASE,
+)
+
+_GATEWAY_PROVIDER_POLICY_RE = re.compile(
+    r"("  # raw provider policy/safety bodies are noisy and may be sensitive
+    r"cybersecurity\s+risk"
+    r"|security\s+policy"
+    r"|safety\s+policy"
+    r"|policy\s+violation"
+    r"|violat(?:e|es|ed|ion)"
+    r"|blocked\s+(?:because|by|under)"
+    r"|request\s+(?:was\s+)?(?:blocked|rejected)"
+    r"|disallowed"
+    r"|moderation"
+    r")",
+    re.IGNORECASE,
+)
+
+_GATEWAY_AUTH_ERROR_RE = re.compile(
+    r"(provider\s+authentication\s+failed|incorrect\s+api\s+key|invalid\s+api\s+key|\b401\b)",
+    re.IGNORECASE,
+)
+
+_GATEWAY_RATE_LIMIT_RE = re.compile(
+    r"(rate\s+limit|rate-limited|\b429\b|quota|usage\s+limit)",
+    re.IGNORECASE,
+)
+
+_GATEWAY_SECRET_PATTERNS = (
+    re.compile(r"\bsk-[A-Za-z0-9][A-Za-z0-9_\-]{12,}\b"),
+    re.compile(r"\bgh[pousr]_[A-Za-z0-9_]{20,}\b"),
+    re.compile(r"\bxox[baprs]-[A-Za-z0-9\-]{20,}\b"),
+    re.compile(r"\bhf_[A-Za-z0-9]{20,}\b"),
+    re.compile(r"\bglpat-[A-Za-z0-9_\-]{20,}\b"),
+    re.compile(r"(?i)\b(Bearer\s+)[A-Za-z0-9._\-]{20,}\b"),
+)
+
+
+def _gateway_platform_value(platform: Any) -> str:
+    """Return a normalized gateway platform value for enums or raw strings."""
+    return str(getattr(platform, "value", platform) or "").strip().lower()
+
+
+def _redact_gateway_user_facing_secrets(text: str) -> str:
+    """Best-effort secret redaction before text can leave the gateway."""
+    redacted = str(text or "")
+    for pattern in _GATEWAY_SECRET_PATTERNS:
+        redacted = pattern.sub(lambda m: (m.group(1) if m.lastindex else "") + "[REDACTED]", redacted)
+    return redacted
+
+
+def _gateway_provider_error_reply(text: str) -> str:
+    """Map raw provider/API errors to a short user-safe Telegram reply."""
+    if _GATEWAY_AUTH_ERROR_RE.search(text):
+        return (
+            "⚠️ Provider authentication failed. Check the configured credentials; "
+            "raw provider details are in the gateway logs."
+        )
+    if _GATEWAY_PROVIDER_POLICY_RE.search(text):
+        return (
+            "⚠️ The model provider rejected the request. I kept the raw provider "
+            "error out of chat; check gateway logs for details or try rephrasing."
+        )
+    if _GATEWAY_RATE_LIMIT_RE.search(text):
+        return "⏱️ The model provider is rate-limiting requests. Please wait a moment and try again."
+    return (
+        "⚠️ The model provider failed after retries. I kept raw provider details "
+        "out of chat; check gateway logs for diagnostics."
+    )
+
+
+def _looks_like_gateway_provider_error(text: str) -> bool:
+    """True when text is infrastructure/provider failure, not normal content."""
+    return bool(_GATEWAY_PROVIDER_ERROR_RE.search(text))
+
+
+def _sanitize_gateway_final_response(platform: Any, text: str) -> str:
+    """Sanitize final gateway replies before sending them to high-noise chats.
+
+    Telegram is Bob's mobile inbox, so it should receive concise, safe provider
+    failure categories instead of raw HTTP bodies, request IDs, or policy text.
+    Other platforms keep the existing behaviour for now.
+    """
+    if not text:
+        return text
+    if _gateway_platform_value(platform) != "telegram":
+        return text
+
+    redacted = _redact_gateway_user_facing_secrets(str(text))
+    if _looks_like_gateway_provider_error(redacted):
+        return _gateway_provider_error_reply(redacted)
+    return redacted
+
+
+def _prepare_gateway_status_message(platform: Any, event_type: str, message: str) -> Optional[str]:
+    """Filter/sanitize agent status callbacks before platform delivery."""
+    text = str(message or "").strip()
+    if not text:
+        return None
+    if _gateway_platform_value(platform) != "telegram":
+        return text
+
+    text = _redact_gateway_user_facing_secrets(text)
+    if _TELEGRAM_NOISY_STATUS_RE.search(text):
+        return None
+    if _looks_like_gateway_provider_error(text):
+        return _gateway_provider_error_reply(text)
+    return text
+
 
 def _telegramize_command_mentions(text: str, platform: Any) -> str:
     """Rewrite slash-command mentions to Telegram-valid command names.
@@ -8248,6 +8386,7 @@ class GatewayRunner:
             response = _normalize_empty_agent_response(
                 agent_result, response, history_len=len(history),
             )
+            response = _sanitize_gateway_final_response(source.platform, response)
 
             # If the agent's session_id changed during compression, update
             # session_entry so transcript writes below go to the right session.
@@ -15782,10 +15921,23 @@ class GatewayRunner:
         def _status_callback_sync(event_type: str, message: str) -> None:
             if not _status_adapter or not _run_still_current():
                 return
+            prepared_message = _prepare_gateway_status_message(
+                source.platform,
+                event_type,
+                message,
+            )
+            if prepared_message is None:
+                logger.debug(
+                    "status_callback suppressed for %s/%s: %s",
+                    source.platform.value if source.platform else "unknown",
+                    event_type,
+                    _redact_gateway_user_facing_secrets(str(message or ""))[:160],
+                )
+                return
             _fut = safe_schedule_threadsafe(
                 _status_adapter.send(
                     _status_chat_id,
-                    message,
+                    prepared_message,
                     metadata=_status_thread_metadata,
                 ),
                 _loop_for_step,