Fix xAI sentiment: use Agent Tools web_search (Live Search deprecated)

xAI returned 410 — search_parameters/Live Search is retired. Route xAI
through the Responses API web_search tool instead (same path as OpenAI):
- OpenAISentimentProvider parametrized with base_url / tool_type / source
- xAI builds it against https://api.x.ai/v1 with the web_search tool
- Drop the dead Live Search code from the generic compatible provider
- Frontend label: "xAI Grok — web search"

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

app/providers/openai_compatible_sentiment.py

@@ -39,21 +39,6 @@ Rules:
 - reasoning should be a brief one-sentence explanation
 """
 
-_SENTIMENT_PROMPT_SEARCH = """\
-Search the web and X for the LATEST news, analyst opinions, and market developments \
-about the stock ticker {ticker} from the past 24-48 hours.
-
-Based on your search findings, analyze the CURRENT market sentiment.
-
-Respond ONLY with a JSON object in this exact format (no markdown, no extra text):
-{{"classification": "<bullish|bearish|neutral>", "confidence": <0-100>, "reasoning": "<brief explanation citing recent news>"}}
-
-Rules:
-- classification must be exactly one of: bullish, bearish, neutral
-- confidence must be an integer from 0 to 100
-- reasoning should cite specific recent news or events you found
-"""
-
 VALID_CLASSIFICATIONS = {"bullish", "bearish", "neutral"}
 
 
@@ -75,8 +60,6 @@ class OpenAICompatibleSentimentProvider:
         model: str,
         base_url: str,
         source: str = "openai_compatible",
-        live_search: bool = False,
-        extra_body: dict | None = None,
     ) -> None:
         if not api_key:
             raise ProviderError("API key is required")
@@ -92,29 +75,8 @@ class OpenAICompatibleSentimentProvider:
         self._client = AsyncOpenAI(api_key=api_key, base_url=base_url, http_client=http_client)
         self._model = model
         self._source = source
-        self._live_search = live_search
-        self._extra_body = extra_body
-
-    @staticmethod
-    def _extract_citations(response: object) -> list[dict[str, str]]:
-        """Best-effort extraction of xAI Live Search citations (list of URLs)."""
-        raw = getattr(response, "citations", None)
-        if not raw:
-            extra = getattr(response, "model_extra", None) or {}
-            raw = extra.get("citations") if isinstance(extra, dict) else None
-        citations: list[dict[str, str]] = []
-        for item in raw or []:
-            if isinstance(item, str):
-                citations.append({"url": item, "title": ""})
-            elif isinstance(item, dict) and item.get("url"):
-                citations.append({"url": str(item["url"]), "title": str(item.get("title", ""))})
-        return citations
 
     async def fetch_sentiment(self, ticker: str) -> SentimentData:
-        prompt = _SENTIMENT_PROMPT_SEARCH if self._live_search else _SENTIMENT_PROMPT
-        kwargs: dict = {}
-        if self._extra_body:
-            kwargs["extra_body"] = self._extra_body
         try:
             response = await self._client.chat.completions.create(
                 model=self._model,
@@ -123,10 +85,9 @@ class OpenAICompatibleSentimentProvider:
                         "role": "system",
                         "content": "You are a financial sentiment analyst. Always respond with valid JSON only, no markdown fences.",
                     },
-                    {"role": "user", "content": prompt.format(ticker=ticker)},
+                    {"role": "user", "content": _SENTIMENT_PROMPT.format(ticker=ticker)},
                 ],
                 temperature=0.3,
-                **kwargs,
             )
             raw_text = (response.choices[0].message.content or "").strip()
             if not raw_text:
@@ -155,7 +116,6 @@ class OpenAICompatibleSentimentProvider:
                 source=self._source,
                 timestamp=datetime.now(timezone.utc),
                 reasoning=reasoning,
-                citations=self._extract_citations(response) if self._live_search else [],
             )
 
         except json.JSONDecodeError as exc:

app/providers/openai_sentiment.py

@@ -55,17 +55,34 @@ VALID_CLASSIFICATIONS = {"bullish", "bearish", "neutral"}
 
 
 class OpenAISentimentProvider:
-    """Fetches sentiment analysis from OpenAI Responses API with live web search."""
+    """Sentiment via the Responses API + web-search tool, with live grounding.
 
-    def __init__(self, api_key: str, model: str = "gpt-4o-mini") -> None:
+    Works against any provider implementing the OpenAI Responses API. OpenAI
+    uses the ``web_search_preview`` tool; xAI Grok uses ``web_search`` at the
+    ``https://api.x.ai/v1`` base URL.
+    """
+
+    def __init__(
+        self,
+        api_key: str,
+        model: str = "gpt-4o-mini",
+        base_url: str | None = None,
+        tool_type: str = "web_search_preview",
+        source: str = "openai",
+    ) -> None:
         if not api_key:
-            raise ProviderError("OpenAI API key is required")
+            raise ProviderError(f"{source} API key is required")
         http_kwargs: dict = {}
         if _CA_BUNDLE and Path(_CA_BUNDLE).exists():
             http_kwargs["verify"] = _CA_BUNDLE
         http_client = httpx.AsyncClient(**http_kwargs)
-        self._client = AsyncOpenAI(api_key=api_key, http_client=http_client)
+        client_kwargs: dict = {"api_key": api_key, "http_client": http_client}
+        if base_url:
+            client_kwargs["base_url"] = base_url
+        self._client = AsyncOpenAI(**client_kwargs)
         self._model = model
+        self._tool_type = tool_type
+        self._source = source
 
     @staticmethod
     def _extract_raw_text(response: object, ticker_context: str) -> str:
@@ -89,12 +106,11 @@ class OpenAISentimentProvider:
             clean = clean[:-3]
         return clean.strip()
 
-    @staticmethod
-    def _normalize_single_result(parsed: dict, ticker: str, citations: list[dict[str, str]]) -> SentimentData:
+    def _normalize_single_result(self, parsed: dict, ticker: str, citations: list[dict[str, str]]) -> SentimentData:
         classification = str(parsed.get("classification", "")).lower()
         if classification not in VALID_CLASSIFICATIONS:
             raise ProviderError(
-                f"Invalid classification '{classification}' from OpenAI for {ticker}"
+                f"Invalid classification '{classification}' from {self._source} for {ticker}"
             )
 
         confidence = int(parsed.get("confidence", 50))
@@ -103,7 +119,8 @@ class OpenAISentimentProvider:
 
         if reasoning:
             logger.info(
-                "OpenAI sentiment for %s: %s (confidence=%d) — %s",
+                "%s sentiment for %s: %s (confidence=%d) — %s",
+                self._source,
                 ticker,
                 classification,
                 confidence,
@@ -114,7 +131,7 @@ class OpenAISentimentProvider:
             ticker=ticker,
             classification=classification,
             confidence=confidence,
-            source="openai",
+            source=self._source,
             timestamp=datetime.now(timezone.utc),
             reasoning=reasoning,
             citations=citations,
@@ -125,7 +142,7 @@ class OpenAISentimentProvider:
         try:
             response = await self._client.responses.create(
                 model=self._model,
-                tools=[{"type": "web_search_preview"}],
+                tools=[{"type": self._tool_type}],
                 instructions="You are a financial sentiment analyst. Always respond with valid JSON only, no markdown fences.",
                 input=_SENTIMENT_PROMPT.format(ticker=ticker),
             )
@@ -172,7 +189,7 @@ class OpenAISentimentProvider:
         try:
             response = await self._client.responses.create(
                 model=self._model,
-                tools=[{"type": "web_search_preview"}],
+                tools=[{"type": self._tool_type}],
                 instructions="You are a financial sentiment analyst. Always respond with valid JSON only, no markdown fences.",
                 input=_SENTIMENT_BATCH_PROMPT.format(tickers_csv=", ".join(normalized)),
             )