aria-code 0.90.0__tar.gz → 0.92.0__tar.gz

{aria_code-0.90.0/src/aria_code.egg-info → aria_code-0.92.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aria-code
-Version: 0.90.0
+Version: 0.92.0
 Summary: Open-source AI agent for coding, financial analysis and logistics operations: numbers you can check, client data kept apart, actions a person approves
 License: Apache-2.0
 Project-URL: Homepage, https://github.com/artheras/aria-code

{aria_code-0.90.0 → aria_code-0.92.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "aria-code"
-version = "0.90.0"
+version = "0.92.0"
 description = "Open-source AI agent for coding, financial analysis and logistics operations: numbers you can check, client data kept apart, actions a person approves"
 readme = "README.md"
 license = { text = "Apache-2.0" }

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/_version.py

@@ -17,4 +17,4 @@ from __future__ import annotations
 
 __all__ = ["__version__"]
 
-__version__ = "0.90.0"
+__version__ = "0.92.0"

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/base.py

@@ -103,6 +103,7 @@ class BaseAgent(ABC):
         self.on_tool_end = on_tool_end
         self.config   = config or {}
         self.lang     = lang
+        self.llm_error: Optional[str] = None   # why the last model call gave nothing, if it failed
         self.memory   = []         # Short-term conversation history for multi-turn context
 
     async def fetch_data(self, symbol: str) -> Dict[str, Any]:
@@ -176,6 +177,12 @@ class BaseAgent(ABC):
         _lang_rule = self._LANG_RULES.get(self.lang, self._LANG_RULES["zh"])
         _data_warn = self._data_guard(quote or {})
         system = system + self._TIME_SENSITIVE_FACT_POLICY + _lang_rule + _data_warn
+        if self.lang == "en":
+            # The agents' own prompts and much of their data are Chinese. A rule
+            # appended at the end lost to them: an English /team synthesis on
+            # Gemini came back in Chinese. Say it first as well.
+            system = ("Write your entire answer in English, even though the instructions and data "
+                      "below may be in Chinese.\n\n" + system)
         from aria_code.providers.llm.base import Message
 
         # 自动注入 DAG pipeline 中的 upstream context
@@ -228,10 +235,14 @@ class BaseAgent(ABC):
                     elif t == "tool_end" and self.on_tool_end:
                         self.on_tool_end(event.get("tool_name", ""), event.get("tool_result", None))
                     elif t == "error":
-                        logger.warning(f"[{self.name}] LLM 错误: {event.get('message')}")
+                        # The caller falls back to a template and says so; a
+                        # WARNING here went to the terminal as a raw log line.
+                        self.llm_error = str(event.get("message") or "error")
+                        logger.info(f"[{self.name}] LLM 错误: {event.get('message')}")
                         break
             except Exception as e:
-                logger.warning(f"[{self.name}] LLM 调用失败: {e}")
+                self.llm_error = str(e) or type(e).__name__
+                logger.info(f"[{self.name}] LLM 调用失败: {e}")
                 break
 
             # 将当前 Assistant 的回复加入消息流

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/earnings.py

@@ -128,9 +128,10 @@ class EarningsAgent(BaseAgent):
         if not earnings:
             return AgentResult(
                 agent=self.name, symbol=symbol,
-                analysis=f"{symbol}: 未获取到财报数据。",
+                analysis=(f"{symbol}: 未获取到财报数据。" if str(self.lang).startswith("zh")
+                          else f"{symbol}: no earnings data."),
                 confidence=0.3, signal="HOLD",
-                key_points=["无财报数据"],
+                key_points=["无财报数据" if str(self.lang).startswith("zh") else "no earnings data"],
             )
 
         earnings_block    = _format_earnings(earnings)
@@ -150,10 +151,10 @@ class EarningsAgent(BaseAgent):
 
         analysis = await self._call_llm(self._SYSTEM, prompt, max_tokens=500)
         if not analysis:
-            analysis = _template_analysis(symbol, earnings, most_recent, beat_miss_summary)
+            analysis = _template_analysis(symbol, earnings, most_recent, beat_miss_summary, lang=self.lang)
 
         signal, confidence = _derive_signal(analysis, most_recent, earnings)
-        key_points         = _build_key_points(earnings, most_recent, signal)
+        key_points         = _build_key_points(earnings, most_recent, signal, lang=self.lang)
 
         return AgentResult(
             agent=self.name, symbol=symbol,
@@ -265,26 +266,30 @@ def _derive_signal(analysis: str, most_recent: Dict, earnings: Dict) -> tuple[st
     return "HOLD", 0.40
 
 
-def _build_key_points(earnings: Dict, most_recent: Dict, signal: str) -> List[str]:
+def _build_key_points(earnings: Dict, most_recent: Dict, signal: str, lang: str = "zh") -> List[str]:
+    zh = str(lang).lower().startswith("zh")
     pts = []
     sup = most_recent.get("pct_surprise")
     if sup is not None:
-        pts.append(f"EPS 超预期 {sup:+.1f}%" if sup >= 0 else f"EPS 低于预期 {sup:.1f}%")
+        if zh:
+            pts.append(f"EPS 超预期 {sup:+.1f}%" if sup >= 0 else f"EPS 低于预期 {sup:.1f}%")
+        else:
+            pts.append(f"EPS beat by {sup:+.1f}%" if sup >= 0 else f"EPS missed by {sup:.1f}%")
     rx = earnings.get("price_reaction_pct")
     if rx is not None:
-        pts.append(f"财报后股价反应: {rx:+.1f}%")
+        pts.append(f"财报后股价反应: {rx:+.1f}%" if zh else f"post-earnings move {rx:+.1f}%")
     rev = earnings.get("revenue_trend", [])
     if len(rev) >= 2:
         v0 = rev[0].get("revenue")
         v1 = rev[1].get("revenue")
         if v0 and v1:
             qoq = (v0 - v1) / abs(v1) * 100
-            pts.append(f"营收环比 {qoq:+.1f}%")
-    pts.append(f"财报信号: {signal}")
+            pts.append(f"营收环比 {qoq:+.1f}%" if zh else f"revenue {qoq:+.1f}% q/q")
+    pts.append(f"财报信号: {signal}" if zh else f"earnings signal: {signal}")
     return pts[:5]
 
 
-def _template_analysis(symbol: str, earnings: Dict, most_recent: Dict, summary: str) -> str:
+def _template_analysis(symbol: str, earnings: Dict, most_recent: Dict, summary: str, lang: str = "zh") -> str:
     sup = most_recent.get("pct_surprise")
     if sup is not None:
         if sup >= 10:   verdict = "STRONG_BEAT"
@@ -295,6 +300,13 @@ def _template_analysis(symbol: str, earnings: Dict, most_recent: Dict, summary:
     else:
         verdict = "IN_LINE"
 
+    if not str(lang).lower().startswith("zh"):
+        return (
+            f"{symbol} earnings (template):\n"
+            f"First read: {summary}\n"
+            f"Rating: {verdict}\n"
+            f"Suggested signal: {'BUY' if 'BEAT' in verdict else ('SELL' if 'MISS' in verdict else 'HOLD')}"
+        )
     return (
         f"{symbol} 财报解读（模板）：\n"
         f"初步判断：{summary}\n"

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/fundamental.py

@@ -109,7 +109,7 @@ class FundamentalAgent(BaseAgent):
 
         analysis  = await self._call_llm(self._SYSTEM, prompt, max_tokens=500, quote=quote)
         if not analysis:
-            analysis = _template_fundamental(symbol, pe, pb, roe, rev_g)
+            analysis = _template_fundamental(symbol, pe, pb, roe, rev_g, lang=self.lang)
 
         signal     = _extract_signal(analysis, pe or 0)
         confidence = _calc_confidence(pe or 0, pb or 0, roe or 0)
@@ -173,7 +173,23 @@ def _extract_key_points(text: str) -> List[str]:
 
 
 def _template_fundamental(symbol: str, pe: Optional[float], pb: Optional[float],
-                           roe: Optional[float], rev_g: Optional[float]) -> str:
+                           roe: Optional[float], rev_g: Optional[float], lang: str = "zh") -> str:
+    if not str(lang).lower().startswith("zh"):
+        if not pe or pe <= 0:
+            valuation, conclusion = "not enough data", "DATA_LIMITED"
+        elif pe < 15:
+            valuation, conclusion = "undervalued", "UNDERVALUED"
+        elif pe > 40:
+            valuation, conclusion = "overvalued", "OVERVALUED"
+        else:
+            valuation, conclusion = "fairly valued", "FAIRLY_VALUED"
+        return (
+            f"{symbol} fundamentals (template):\n"
+            f"• Valuation: PE={_fmt_num(pe, 1, 'x')}  PB={_fmt_num(pb, 2, 'x')}  → {valuation}\n"
+            f"• Profitability: ROE={_fmt_num(roe, 1, '%')}  revenue growth={_fmt_num(rev_g, 1, '%')}\n"
+            "• Missing figures are not zeros; check against the filings and peers\n"
+            f"• Conclusion: {conclusion}"
+        )
     if not pe or pe <= 0:
         valuation, conclusion = "数据不足", "DATA_LIMITED"
     elif pe < 15:

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/macro.py

@@ -54,7 +54,7 @@ class MacroAgent(BaseAgent):
 
         analysis = await self._call_llm(self._SYSTEM, prompt, max_tokens=500, quote=quote)
         if not analysis:
-            analysis = _template_macro(symbol, indices)
+            analysis = _template_macro(symbol, indices, lang=self.lang)
 
         signal     = _extract_signal(analysis)
         key_points = _extract_key_points(analysis)
@@ -85,15 +85,27 @@ def _extract_key_points(text: str) -> List[str]:
 
 _IDX_NAMES = {"000001": "上证", "399006": "创业板"}
 
-def _template_macro(symbol: str, indices: Dict) -> str:
-    idx_str = "、".join(
-        f"{_IDX_NAMES.get(k, k)} {v.get('change_pct', 0):+.2f}%"
+def _template_macro(symbol: str, indices: Dict, lang: str = "zh") -> str:
+    """Without a model: the index moves, and nothing claimed beyond them.
+
+    This stated "当前货币政策偏宽松，流动性充裕" (policy loose, liquidity
+    ample) as fact — a macro view with no data behind it.
+    """
+    zh = str(lang).lower().startswith("zh")
+    idx_str = ("、" if zh else ", ").join(
+        f"{_IDX_NAMES.get(k, k) if zh else k} {v.get('change_pct', 0):+.2f}%"
         for k, v in indices.items()
-    ) if indices else "指数数据不可用"
+    ) if indices else ("指数数据不可用" if zh else "index data unavailable")
+    if not zh:
+        return (
+            f"{symbol} macro (template):\n"
+            f"• Market indices: {idx_str}\n"
+            "• Policy and liquidity were not assessed (no model analysis)\n"
+            "• Conclusion: NEUTRAL"
+        )
     return (
         f"{symbol} 宏观分析（模板）:\n"
         f"• 市场指数: {idx_str}\n"
-        "• 当前货币政策偏宽松，流动性充裕\n"
-        "• 建议关注政策面变化和外资动向\n"
+        "• 未评估货币政策与流动性（无模型分析）\n"
         "• 结论: NEUTRAL"
     )

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/news.py

@@ -90,9 +90,10 @@ class NewsAgent(BaseAgent):
         if not news:
             return AgentResult(
                 agent=self.name, symbol=symbol,
-                analysis=f"{symbol}: 未获取到近期新闻数据。",
+                analysis=(f"{symbol}: 未获取到近期新闻数据。" if str(self.lang).startswith("zh")
+                          else f"{symbol}: no recent news."),
                 confidence=0.3, signal="HOLD",
-                key_points=["无近期新闻数据"],
+                key_points=["无近期新闻数据" if str(self.lang).startswith("zh") else "no recent news"],
             )
 
         events = _classify_events(news)
@@ -110,11 +111,11 @@ class NewsAgent(BaseAgent):
 
         analysis = await self._call_llm(self._SYSTEM, prompt, max_tokens=500)
         if not analysis:
-            analysis = _template_analysis(symbol, news, events)
+            analysis = _template_analysis(symbol, news, events, lang=self.lang)
 
         signal     = _extract_signal(analysis, events)
         confidence = _estimate_confidence(news, events)
-        key_points = _build_key_points(news, events)
+        key_points = _build_key_points(news, events, lang=self.lang)
 
         return AgentResult(
             agent=self.name, symbol=symbol,
@@ -170,12 +171,21 @@ def _estimate_confidence(news: List[Dict], events: Dict[str, int]) -> float:
     return min(round(base, 2), 0.75)
 
 
-def _build_key_points(news: List[Dict], events: Dict[str, int]) -> List[str]:
+_EVENT_EN = {"earnings": "earnings-related", "upgrade": "analyst upgrades", "downgrade": "analyst downgrades",
+             "insider": "insider trades", "regulatory": "regulatory events", "dividend": "dividend news",
+             "merger": "M&A news"}
+
+
+def _build_key_points(news: List[Dict], events: Dict[str, int], lang: str = "zh") -> List[str]:
+    zh = str(lang).lower().startswith("zh")
     points = []
     recent = [n for n in news if n.get("age_days", 999) <= 3]
     if recent:
-        points.append(f"近3日 {len(recent)} 条新鲜新闻")
+        points.append(f"近3日 {len(recent)} 条新鲜新闻" if zh else f"{len(recent)} stories in the last 3 days")
     for etype, count in events.items():
+        if not zh:
+            points.append(f"{_EVENT_EN.get(etype, etype)} × {count}")
+            continue
         label = {
             "earnings": "业绩/财报相关",
             "upgrade": "分析师升级评级",
@@ -189,8 +199,22 @@ def _build_key_points(news: List[Dict], events: Dict[str, int]) -> List[str]:
     return points[:5]
 
 
-def _template_analysis(symbol: str, news: List[Dict], events: Dict[str, int]) -> str:
+def _template_analysis(symbol: str, news: List[Dict], events: Dict[str, int], lang: str = "zh") -> str:
     recent = [n for n in news if n.get("age_days", 999) <= 7]
+    if not str(lang).lower().startswith("zh"):
+        tone = "neutral"
+        if events.get("upgrade", 0) > events.get("downgrade", 0):
+            tone = "positive"
+        elif events.get("downgrade", 0) > 0 or events.get("regulatory", 0) > 0:
+            tone = "negative"
+        titles = "\n".join(f"  • {n['title'][:60]}" for n in recent[:3])
+        return (
+            f"{symbol} recent news tone: {tone}\n"
+            f"{len(recent)} related stories in the last 7 days\n"
+            f"Headlines:\n{titles or '  (no recent headlines)'}\n"
+            f"Events: {', '.join(events.keys()) or 'none'}\n"
+            f"Conclusion: {tone.upper() if tone != 'neutral' else 'NEUTRAL'}"
+        )
     sentiment = "中性"
     if events.get("upgrade", 0) > events.get("downgrade", 0):
         sentiment = "偏正面"

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/risk.py

@@ -94,15 +94,16 @@ class RiskAgent(BaseAgent):
 
         analysis  = await self._call_llm(self._SYSTEM, prompt, max_tokens=400, quote=quote)
         if not analysis:
-            analysis = _template_risk(symbol, metrics)
+            analysis = _template_risk(symbol, metrics, lang=self.lang)
 
         signal     = _risk_to_signal(metrics)
         confidence = 0.70
+        zh = str(self.lang).lower().startswith("zh")
         key_points = [
-            f"年化波动率 {metrics.get('ann_vol',0):.1f}%",
-            f"最大回撤 {metrics.get('max_dd',0):.1f}%",
-            f"夏普比率 {metrics.get('sharpe',0):.2f}",
-        ] if metrics else ["风险数据不可用"]
+            f"年化波动率 {metrics.get('ann_vol',0):.1f}%" if zh else f"annualised volatility {metrics.get('ann_vol',0):.1f}%",
+            f"最大回撤 {metrics.get('max_dd',0):.1f}%" if zh else f"max drawdown {metrics.get('max_dd',0):.1f}%",
+            f"夏普比率 {metrics.get('sharpe',0):.2f}" if zh else f"Sharpe ratio {metrics.get('sharpe',0):.2f}",
+        ] if metrics else ["风险数据不可用" if zh else "risk data unavailable"]
 
         return AgentResult(
             agent=self.name, symbol=symbol,
@@ -151,9 +152,19 @@ def _risk_to_signal(metrics: Dict) -> str:
     return "HOLD"
 
 
-def _template_risk(symbol: str, m: Dict) -> str:
+def _template_risk(symbol: str, m: Dict, lang: str = "zh") -> str:
     vol = m.get("ann_vol", 0)
     dd  = m.get("max_dd",  0)
+    if not str(lang).lower().startswith("zh"):
+        level = "high" if vol > 40 else ("medium" if vol > 20 else "low")
+        return (
+            f"{symbol} risk (template):\n"
+            f"• Annualised volatility: {vol:.1f}% (risk {level})\n"
+            f"• Max drawdown: {dd:.1f}%\n"
+            f"• Sharpe ratio: {m.get('sharpe',0):.2f}\n"
+            f"• Suggested position: {'5-10%' if vol>40 else '10-15%' if vol>20 else '15-20%'}\n"
+            f"• Risk Score: {min(10, int(vol/5))}/10"
+        )
     risk_level = "高" if vol > 40 else ("中" if vol > 20 else "低")
     return (
         f"{symbol} 风险分析（模板）:\n"

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/financial/technical.py

@@ -58,11 +58,11 @@ class TechnicalAgent(BaseAgent):
 
         analysis = await self._call_llm(self._SYSTEM, prompt, max_tokens=600, quote=quote)
         if not analysis:
-            analysis = _template_analysis(symbol, price, history)
+            analysis = _template_analysis(symbol, price, history, lang=self.lang)
 
         signal     = _extract_signal(analysis, history)
         confidence = history.get("signal_strength", 0.5)
-        key_points = _extract_key_points(history, price)
+        key_points = _extract_key_points(history, price, lang=self.lang)
 
         return AgentResult(
             agent=self.name, symbol=symbol,
@@ -218,35 +218,54 @@ def _extract_signal(analysis: str, history: Dict) -> str:
     return "HOLD"
 
 
-def _extract_key_points(history: Dict, price: float) -> List[str]:
+# Pattern values stay as _detect_simple_pattern returns them; this is how
+# they read in English.
+_PATTERN_EN = {"锤子线": "hammer", "阳线吞噬": "bullish engulfing", "十字星": "doji", "数据不足": "not enough data"}
+
+
+def _extract_key_points(history: Dict, price: float, lang: str = "zh") -> List[str]:
+    """The analyst's one-line findings, in the team's language.
+
+    They were Chinese only, so an English /team listed "距MA20 -0.5%" and
+    "MACD死叉，空头压力" under "Technical".
+    """
+    zh = str(lang).lower().startswith("zh")
     points = []
     rsi = history.get("rsi", 50)
     ma20 = history.get("ma20", 0)
-    ma60 = history.get("ma60", 0)
     if rsi < 35:
-        points.append(f"RSI超卖({rsi:.0f})，有反弹机会")
+        points.append(f"RSI超卖({rsi:.0f})，有反弹机会" if zh else f"RSI oversold ({rsi:.0f}), room to rebound")
     elif rsi > 70:
-        points.append(f"RSI超买({rsi:.0f})，注意回调风险")
+        points.append(f"RSI超买({rsi:.0f})，注意回调风险" if zh else f"RSI overbought ({rsi:.0f}), pullback risk")
     if ma20 > 0:
         diff_pct = (price - ma20) / ma20 * 100
-        points.append(f"距MA20 {diff_pct:+.1f}%")
+        points.append(f"距MA20 {diff_pct:+.1f}%" if zh else f"{diff_pct:+.1f}% vs MA20")
     macd = history.get("macd", 0)
     sig  = history.get("macd_signal", 0)
     if macd > sig and history.get("macd_hist", 0) > 0:
-        points.append("MACD金叉，多头动能")
+        points.append("MACD金叉，多头动能" if zh else "MACD bullish crossover")
     elif macd < sig:
-        points.append("MACD死叉，空头压力")
+        points.append("MACD死叉，空头压力" if zh else "MACD bearish crossover")
     pattern = history.get("pattern", "")
     if pattern and pattern != "无特殊形态":
-        points.append(f"K线形态: {pattern}")
+        points.append(f"K线形态: {pattern}" if zh else f"Candle pattern: {_PATTERN_EN.get(pattern, pattern)}")
     return points
 
 
-def _template_analysis(symbol: str, price: float, history: Dict) -> str:
+def _template_analysis(symbol: str, price: float, history: Dict, lang: str = "zh") -> str:
     rsi  = history.get("rsi", 50)
     ma20 = history.get("ma20", 0)
     macd = history.get("macd", 0)
     sig  = history.get("macd_signal", 0)
+    if not str(lang).lower().startswith("zh"):
+        up, bull = price > ma20, macd > sig
+        return (
+            f"{symbol} technicals (template):\n"
+            f"• Price {price}, {'uptrend' if up else 'downtrend'} (MA20={ma20:.2f})\n"
+            f"• RSI={rsi:.0f}, {'oversold' if rsi < 35 else 'overbought' if rsi > 70 else 'normal range'}\n"
+            f"• MACD momentum {'bullish' if bull else 'bearish'}\n"
+            f"• Overall {'BULLISH' if bull and up else 'NEUTRAL'}"
+        )
     trend = "上升趋势" if price > ma20 else "下降趋势"
     momentum = "偏多" if macd > sig else "偏空"
     return (

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/portfolio_agent.py

@@ -176,20 +176,24 @@ class PortfolioAgent(BaseAgent):
     async def analyze_portfolio(
         self, symbols: List[str], data: Dict[str, Any]
     ) -> AgentResult:
+        zh = self.lang != "en"
         if len(symbols) < 2:
             return AgentResult(
                 agent=self.name, symbol=",".join(symbols),
-                analysis="组合分析需要至少 2 个标的。",
+                analysis=("组合分析需要至少 2 个标的。" if zh
+                          else "A portfolio analysis needs at least 2 symbols."),
                 confidence=0.0, signal="HOLD",
-                key_points=["标的数量不足"],
+                key_points=["标的数量不足" if zh else "Not enough symbols"],
             )
 
         if data.get("error") == "insufficient_data":
             return AgentResult(
                 agent=self.name, symbol=",".join(symbols),
-                analysis="无法获取足够的历史价格数据进行组合分析。",
+                analysis=("无法获取足够的历史价格数据进行组合分析。" if zh
+                          else "Not enough price history could be fetched to analyse the portfolio."),
                 confidence=0.3, signal="HOLD",
-                key_points=["历史数据不足（需要 >20 个交易日）"],
+                key_points=["历史数据不足（需要 >20 个交易日）" if zh
+                            else "Not enough history (needs more than 20 trading days)"],
             )
 
         valid_syms = data.get("valid_symbols", symbols)
@@ -210,12 +214,12 @@ class PortfolioAgent(BaseAgent):
 
         analysis = await self._call_llm(self._SYSTEM, prompt, max_tokens=650)
         if not analysis:
-            analysis = _template_analysis(valid_syms, data)
+            analysis = _template_analysis(valid_syms, data, self.lang)
 
         verdict    = _extract_verdict(analysis)
         signal     = _verdict_to_signal(verdict)
         confidence = _estimate_confidence(data)
-        key_points = _build_key_points(data, verdict)
+        key_points = _build_key_points(data, verdict, self.lang)
 
         return AgentResult(
             agent=self.name,
@@ -225,6 +229,9 @@ class PortfolioAgent(BaseAgent):
             signal=signal,
             key_points=key_points,
             data_used={
+                # The verdict itself: the signal maps NEEDS_ATTENTION to SELL,
+                # and a banner built from the signal said HIGH_RISK.
+                "verdict": verdict,
                 "n": len(valid_syms),
                 "port_vol_ann": data.get("port_vol_ann"),
                 "div_ratio": data.get("div_ratio"),
@@ -312,35 +319,45 @@ def _estimate_confidence(d: Dict) -> float:
     return min(round(base, 2), 0.75)
 
 
-def _build_key_points(d: Dict, verdict: str) -> List[str]:
+def _build_key_points(d: Dict, verdict: str, lang: str = "zh") -> List[str]:
+    zh  = lang != "en"
     pts = []
     n   = len(d.get("valid_symbols", []))
     if d.get("weight_source") == "ledger":
-        pts.append(f"{n} 个标的，按真实持仓成本加权")
+        pts.append(f"{n} 个标的，按真实持仓成本加权" if zh
+                   else f"{n} symbols, weighted by the cost of your real positions")
     else:
-        pts.append(f"{n} 个标的，等权配置（无真实持仓数据，非实际风险暴露）")
+        pts.append(f"{n} 个标的，等权配置（无真实持仓数据，非实际风险暴露）" if zh
+                   else f"{n} symbols, equal weight (no real positions recorded, so not your actual exposure)")
 
     port_vol = d.get("port_vol_ann", 0)
     if port_vol:
-        risk_lv = "高风险" if port_vol > 0.30 else ("中等" if port_vol > 0.18 else "低风险")
-        pts.append(f"组合年化波动 {port_vol*100:.1f}%（{risk_lv}）")
+        if zh:
+            risk_lv = "高风险" if port_vol > 0.30 else ("中等" if port_vol > 0.18 else "低风险")
+            pts.append(f"组合年化波动 {port_vol*100:.1f}%（{risk_lv}）")
+        else:
+            risk_lv = "high" if port_vol > 0.30 else ("medium" if port_vol > 0.18 else "low")
+            pts.append(f"Annualised portfolio volatility {port_vol*100:.1f}% ({risk_lv})")
 
     div_r = d.get("div_ratio", 1)
     if div_r < 1.1:
-        pts.append(f"分散度不足（ratio {div_r:.2f}x，标的相关性高）")
+        pts.append(f"分散度不足（ratio {div_r:.2f}x，标的相关性高）" if zh
+                   else f"Poorly diversified (ratio {div_r:.2f}x, the symbols move together)")
     else:
-        pts.append(f"分散度合理（ratio {div_r:.2f}x）")
+        pts.append(f"分散度合理（ratio {div_r:.2f}x）" if zh
+                   else f"Reasonably diversified (ratio {div_r:.2f}x)")
 
     high_corr = d.get("high_corr", [])
     if high_corr:
         top = high_corr[0]
-        pts.append(f"最高相关对: {top['sym1']}↔{top['sym2']} ({top['corr']:+.2f})")
+        pts.append(f"{'最高相关对' if zh else 'Most correlated pair'}: "
+                   f"{top['sym1']}↔{top['sym2']} ({top['corr']:+.2f})")
 
-    pts.append(f"整体评级: {verdict}")
+    pts.append(f"{'整体评级' if zh else 'Overall'}: {verdict}")
     return pts[:6]
 
 
-def _template_analysis(symbols: List[str], d: Dict) -> str:
+def _template_analysis(symbols: List[str], d: Dict, lang: str = "zh") -> str:
     high_corr = d.get("high_corr", [])
     port_vol  = d.get("port_vol_ann", 0)
     div_r     = d.get("div_ratio", 1)
@@ -351,6 +368,18 @@ def _template_analysis(symbols: List[str], d: Dict) -> str:
     elif len(high_corr) >= 3 or port_vol > 0.30:
         verdict = "HIGH_RISK"
 
+    if lang == "en":
+        lines = [f"{','.join(symbols)} portfolio report (template)"]
+        lines.append(f"Annualised volatility: {port_vol*100:.1f}%")
+        lines.append(f"Diversification ratio: {div_r:.2f}x")
+        if high_corr:
+            lines.append(f"{len(high_corr)} highly correlated pairs, the highest "
+                         f"{high_corr[0]['sym1']}↔{high_corr[0]['sym2']} {high_corr[0]['corr']:+.2f}")
+        else:
+            lines.append("No highly correlated pairs.")
+        lines.append(f"\n{verdict}")
+        return "\n".join(lines)
+
     lines = [f"{','.join(symbols)} 组合分析报告（模板）"]
     lines.append(f"年化波动率: {port_vol*100:.1f}%")
     lines.append(f"分散度系数: {div_r:.2f}x")

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/registry.py

@@ -30,9 +30,19 @@ _BUILTIN: Dict[str, Type[BaseAgent]] = {
 }
 
 # 尝试加载其他内置 agent（文件存在则注册）
+# The package this module was loaded as: "aria_code.agents", or "agents" when the
+# CLI's bare import root is in use. The built-ins were imported by the fixed
+# name "aria_code.agents.…" — a second copy of the package under the other
+# root, whose classes subclass a different BaseAgent. issubclass() then said
+# no and they were dropped without a word: /team announced four analysts and
+# only "technical" (imported relatively, above) ever ran.
+_PACKAGE = __package__ or "aria_code.agents"
+
+
 def _try_import_builtin(module_path: str, name: str) -> None:
+    if module_path.startswith("aria_code.agents."):
+        module_path = _PACKAGE + module_path[len("aria_code.agents"):]
     try:
-        parts = module_path.split(".")
         mod = importlib.import_module(module_path)
         for _, cls in inspect.getmembers(mod, inspect.isclass):
             if issubclass(cls, BaseAgent) and cls is not BaseAgent:

{aria_code-0.90.0 → aria_code-0.92.0}/src/aria_code/agents/team.py

@@ -316,6 +316,7 @@ class AgentTeam:
                     llm_provider=self.llm,
                     data_router=self.data,
                     on_token=self.on_token,
+                    lang=self.lang,   # without it the agent answered in Chinese for an English /team
                 )
                 synth_data = {
                     "agent_results": [r.to_dict() for r in results],
@@ -332,11 +333,11 @@ class AgentTeam:
                     synthesis_text = synth_result.analysis
                 except Exception as e:
                     logger.warning(f"[synthesis] 失败: {e}")
-                    synthesis_text = _template_synthesis(results, self.signal_scheme)
+                    synthesis_text = _template_synthesis(results, self.signal_scheme, lang=self.lang)
             else:
-                synthesis_text = _template_synthesis(results, self.signal_scheme)
+                synthesis_text = _template_synthesis(results, self.signal_scheme, lang=self.lang)
         else:
-            synthesis_text = _template_synthesis(results, self.signal_scheme)
+            synthesis_text = _template_synthesis(results, self.signal_scheme, lang=self.lang)
 
         return TeamResult(
             symbol       = symbol,
@@ -386,7 +387,7 @@ class AgentTeam:
 
         # 执行最终的信号表决与综合
         final_signal, confidence = self.signal_scheme.vote(results)
-        synthesis_text = _template_synthesis(results, self.signal_scheme)
+        synthesis_text = _template_synthesis(results, self.signal_scheme, lang=self.lang)
 
         return TeamResult(
             symbol       = symbol,
@@ -468,25 +469,32 @@ def _vote_signal(results: List[AgentResult], scheme: SignalScheme = FINANCIAL_SC
     return scheme.vote(results)
 
 
-def _template_synthesis(results: List[AgentResult], scheme: SignalScheme = FINANCIAL_SCHEME) -> str:
-    """无 synthesis agent 时的模板汇总"""
+def _template_synthesis(results: List[AgentResult], scheme: SignalScheme = FINANCIAL_SCHEME,
+                        lang: str = "zh") -> str:
+    """无 synthesis agent 时的模板汇总 — the summary when no model wrote one, in the team's language."""
+    zh = str(lang).lower().startswith("zh")
+    T = (lambda a, b: a) if zh else (lambda a, b: b)
     if not results:
-        return "分析完成，无结果。"
-    lines = ["## 团队分析汇总\n"]
+        return T("分析完成，无结果。", "Analysis finished with no results.")
+    lines = [T("## 团队分析汇总\n", "## Team summary\n")]
     failed_count = sum(1 for r in results if not r.success)
     if failed_count:
-        lines.append(f"> ⚠️ {failed_count}/{len(results)} 个 agent 未能完成分析"
-                     f"（超时或 LLM 不可用），以下结论仅基于成功的 agent。\n")
+        lines.append(T(f"> ⚠️ {failed_count}/{len(results)} 个 agent 未能完成分析"
+                       f"（超时或 LLM 不可用），以下结论仅基于成功的 agent。\n",
+                       f"> ⚠️ {failed_count} of {len(results)} analysts did not finish (timeout or no model); "
+                       "the conclusion uses only the ones that did.\n"))
     for r in results:
         if r.success:
-            lines.append(f"**{r.agent.upper()}** ({r.signal}, 置信度 {r.confidence:.0%})")
+            lines.append(f"**{r.agent.upper()}** ({r.signal}, {T('置信度', 'confidence')} {r.confidence:.0%})")
             for pt in (r.key_points or [])[:3]:
                 lines.append(f"  • {pt}")
         else:
-            err_label = "超时" if r.error == "timeout" else (r.error or "分析失败")
+            err_label = T("超时", "timed out") if r.error == "timeout" else (r.error or T("分析失败", "failed"))
             lines.append(f"**{r.agent.upper()}** ⚠️ {err_label}")
     signal, conf = scheme.vote(results)
-    lines.append(f"\n**综合结论**: {signal}（置信度 {conf:.0%}）")
+    lines.append(f"\n**{T('综合结论', 'Conclusion')}**: {signal}（{T('置信度', 'confidence')} {conf:.0%}）"
+                 if zh else f"\n**Conclusion**: {signal} (confidence {conf:.0%})")
     if failed_count == len(results):
-        lines.append("\n> ⚠️ 所有 agent 均未成功，此结论仅为默认值，不具参考意义。请确认 LLM 服务正常后重试。")
+        lines.append(T("\n> ⚠️ 所有 agent 均未成功，此结论仅为默认值，不具参考意义。请确认 LLM 服务正常后重试。",
+                       "\n> ⚠️ No analyst succeeded; this is a default, not a conclusion. Check the model and retry."))
     return "\n".join(lines)