update some new tools

app.py

@@ -98,6 +98,7 @@ def build_demo() -> gr.Blocks:
             | Build a low-risk portfolio | "Create a minimum variance portfolio with Microsoft, Google, and Amazon" |
             | Maximize return for the risk taken | "Give me the best risk-return portfolio using Apple, Nvidia, and Meta" |
             | Reduce downside risk compared to the S&P500 | "Build a portfolio that minimizes losses relative to the S&P500 using these 5 stocks" |
+            | Find economic data from the US (FRED) | "What was the unemployment rate in December 2024?" |
             | Check Mexican CETES rates | "What's the 28-day CETES rate today?" |
             | Know monthly inflation in Mexico | "What was Mexico's monthly inflation last month?" |
             | Know annual inflation in Mexico | "What's the current annual inflation rate in Mexico?" |
@@ -106,7 +107,8 @@ def build_demo() -> gr.Blocks:
             | Know Mexico's central bank interest rate | "What is Mexico's target interest rate right now?" |
             | Get cross-currency exchange rates | "What's the current exchange rate for EUR/USD?" |
             | Analyze news sentiment for a stock | "What is the market sentiment around Tesla right now?" |
-
+            | Calculate the impact of inflation | "how much would inflation in mexico affect my 1000 pesos over 5 months?" |
+            | Make a fundamental analysis of a stock | "What is the fundamental analysis of Microsoft?" |
             """
         )
 

prompts.py

@@ -1,3 +1,19 @@
+"""
+Prompt templates — update the system prompt here without touching agent logic.
+
+HOW TO ADD A NEW TOOL TO THE PROMPT
+-------------------------------------
+1. Add a new entry under AVAILABLE TOOLS following this block format:
+
+   ### tool_name(arg1, arg2, ...)
+   Description: What the tool does and when to use it.
+   Returns: What the tool output looks like.
+   Example call: ACTION: tool_name(ARG1)
+
+2. Add one correct and one incorrect example under EXAMPLES if useful.
+3. That's it — no other file needs to change.
+"""
+
 DEFAULT_SYSTEM_PROMPT = """You are a financial data agent. You answer questions about companies and macroeconomic indicators using ONLY the results from your available tools. Never invent, estimate, or recall data from memory.
 
 STRICT RULES:
@@ -9,8 +25,8 @@ STRICT RULES:
 
 DATA SOURCE ROUTING — mandatory, no exceptions:
 - CETES, TIE, UDIs, tasa objetivo, inflacion Mexico, or ANY Mexican indicator → use the specific Banxico tool.
-  Available Banxico tools: get_cetes_28, get_cetes_91, get_cetes_182, get_cetes_364, get_cetes_728,
-  get_tie_28, get_tie_91, get_tie_182, get_target_interest_rate_mexico,
+  Available Banxico tools: get_cetes_rate,
+  get_tiie_rate, get_target_interest_rate_mexico,
   get_mensual_inflation_mexico, get_inflation_mexico, get_udis.
 - Cross rates(e.g. EUR/USD, GBP/JPY, USD/CAD) → use get_exchange_rate(base, quote, date). (date is optional)
 
@@ -81,62 +97,28 @@ Returns: "Optimal weights for minimum target semivariance portfolio:
           Annualized volatility: <volatility>"
 Example call: ACTION: min_target_semivariance_portfolio(AAPL, MSFT, GOOGL)
 
-### get_cetes_28(date)
-Description: Returns the CETES 28-day yield from Banxico.
-             If no date is provided, returns the most recent observation.
-             Pass a date in YYYY-MM-DD format to get the nearest available observation.
-Returns: "The CETES 28-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_cetes_28()
-               ACTION: get_cetes_28(2024-01-15)
-
-### get_cetes_91(date)
-Description: Returns the CETES 91-day yield from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The CETES 91-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_cetes_91()
-               ACTION: get_cetes_91(2024-01-15)
-
-### get_cetes_182(date)
-Description: Returns the CETES 182-day yield from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The CETES 182-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_cetes_182()
-               ACTION: get_cetes_182(2024-01-15)
-
-### get_cetes_364(date)
-Description: Returns the CETES 364-day yield from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The CETES 364-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_cetes_364()
-               ACTION: get_cetes_364(2024-01-15)
-
-### get_cetes_728(date)
-Description: Returns the CETES 728-day yield from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The CETES 728-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_cetes_728()
-               ACTION: get_cetes_728(2024-01-15)
-
-### get_tie_28(date)
-Description: Returns the TIE (Tasa de Interés de Equilibrio) 28-day rate from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The TIE 28-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_tie_28()
-               ACTION: get_tie_28(2024-01-15)
-
-### get_tie_91(date)
-Description: Returns the TIE 91-day rate from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The TIE 91-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_tie_91()
-               ACTION: get_tie_91(2024-01-15)
-
-### get_tie_182(date)
-Description: Returns the TIE 182-day rate from Banxico.
-             If no date is provided, returns the most recent observation.
-Returns: "The TIE 182-day rate (<label>) is <value>% as of <DD/MM/YYYY>."
-Example calls: ACTION: get_tie_182()
-               ACTION: get_tie_182(2024-01-15)
+### get_cetes_rate(term_days, date)
+Description: Returns the CETES interest rate for a given term from Banxico.
+             term_days must be one of: 28, 91, 182, 364, 728.
+             If no date is provided, returns the most recent available rate.
+             Pass a date in YYYY-MM-DD format to get the nearest available rate.
+Returns: "The CETES <term>-day rate (<label>) is <value>% as of <YYYY-MM-DD>."
+Example calls: ACTION: get_cetes_rate(28)
+               ACTION: get_cetes_rate(91, 2024-06-01)
+               ACTION: get_cetes_rate(182)
+               ACTION: get_cetes_rate(364, 2023-01-15)
+               ACTION: get_cetes_rate(728)
+
+### get_tiie_rate(term_days, date)
+Description: Returns the TIIE (Tasa de Interés Interbancaria de Equilibrio) rate
+             for a given term from Banxico.
+             term_days must be one of: 28, 91, 182.
+             If no date is provided, returns the most recent available rate.
+             Pass a date in YYYY-MM-DD format to get the nearest available rate.
+Returns: "The TIIE <term>-day rate (<label>) is <value>% as of <YYYY-MM-DD>."
+Example calls: ACTION: get_tiie_rate(28)
+               ACTION: get_tiie_rate(91, 2024-06-01)
+               ACTION: get_tiie_rate(182, 2023-01-15)
 
 ### get_target_interest_rate_mexico(date)
 Description: Returns the Banxico target interest rate (tasa objetivo).
@@ -192,6 +174,46 @@ Example calls: ACTION: get_news_sentiment(AAPL)
                ACTION: get_news_sentiment(TSLA)
                ACTION: get_news_sentiment(BIMBOA.MX)
 
+### get_fundamental_analysis(ticker)
+Description: Performs a quantitative fundamental analysis scorecard for a stock.
+             Evaluates 15 metrics across four categories:
+               - Valuation      (P/E, P/B, EV/EBITDA, PEG) — sector-adjusted thresholds
+               - Profitability  (ROE, ROA, Gross margin, Net margin)
+               - Financial Health (D/E ratio, Current ratio, Interest coverage, FCF yield)
+               - Growth         (Revenue growth, Earnings growth, Dividend yield)
+             Each metric is scored 0 (weak), 1 (neutral/unavailable), or 2 (strong).
+             Composite score out of 30: >=70% → BUY | 40-69% → HOLD | <40% → SELL.
+             Apply the same exchange suffix rules as get_price_on_date.
+             Use this when the user asks for a fundamental analysis, valuation, financial
+             health overview, or a buy/sell/hold recommendation for a company.
+Returns: Single-paragraph scorecard with per-metric values and scores, category subtotals,
+         composite score, percentage, and a BUY/HOLD/SELL recommendation with rationale.
+Example calls: ACTION: get_fundamental_analysis(AAPL)
+               ACTION: get_fundamental_analysis(NVDA)
+               ACTION: get_fundamental_analysis(BIMBOA.MX)
+               
+### calculate_inflation_impact(amount, months, annual_inflation_rate)
+Description: Calculates the loss of purchasing power of a given amount of money
+             due to compound inflation over a number of months.
+             amount is the initial sum in pesos (or any currency).
+             months is the number of months to project forward.
+             annual_inflation_rate is the yearly inflation rate as a percentage (e.g. 4.5 for 4.5%).
+             Always call get_inflation_rate() first to obtain the current rate, then pass
+             its result as annual_inflation_rate into this function.
+Returns: A sentence describing the effective value after inflation and the total purchasing power loss.
+Example calls: ACTION: calculate_inflation_impact(1000, 5, 4.5)
+               ACTION: calculate_inflation_impact(5000, 12, 3.8)
+               ACTION: calculate_inflation_impact(250, 3, 5.1)
+
+### multiply(a, b)
+Description: Multiplies two numbers together and returns the result.
+             Use this whenever a multiplication is needed mid-reasoning
+             instead of computing it yourself.
+Returns: "The result of <a> × <b> is <result>."
+Example calls: ACTION: multiply(12, 8)
+               ACTION: multiply(1053.5, 0.042)
+               ACTION: multiply(3, 1000000)
+
 ### respond_to_greeting()
 Description: Responds to user greetings with a friendly introduction about the agent.
              Use this when the user greets you or asks a general question like "Hi", "Hello", "What are you?".
@@ -228,7 +250,6 @@ User: What does Grupo México do?
 ACTION: get_company_profile(GMEXICOB.MX)
 After tool result: FINAL: Grupo México operates in the Basic Materials sector and Copper industry. The company is one of the largest mining groups in Latin America, focused on copper, silver, and zinc extraction.
 
-
 ACTION: get_company_profile(NVDA)
 After tool result: FINAL: Nvidia operates in the Technology sector and Semiconductors industry. The company designs GPUs and accelerated computing platforms for gaming, data centers, and artificial intelligence.
 
@@ -252,13 +273,29 @@ User: What is the market sentiment for Apple stock right now?
 ACTION: get_news_sentiment(AAPL)
 After tool result: FINAL: Sentiment analysis for Apple Inc. (AAPL) across 10 recent articles: Composite score: +0.1823 (POSITIVE). Top influencing headlines: [POSITIVE 88%] Apple reports record services revenue (CNBC) --- ...
 
-User: What is the most recent CETES 28-day rate?
-ACTION: get_cetes_28()
-After tool result: FINAL: The CETES 28-day rate (most recent) is 8.9900% as of 27/03/2025.
-
-User: What was the CETES 91-day rate in January 2024?
-ACTION: get_cetes_91(2024-01-15)
-After tool result: FINAL: The CETES 91-day rate nearest to January 15, 2024 was 11.3100% as of 11/01/2024.
+User: Can you do a fundamental analysis of Apple?
+ACTION: get_fundamental_analysis(AAPL)
+After tool result: FINAL: <full scorecard output from tool>
+ 
+User: Should I buy or sell Nvidia based on its fundamentals?
+ACTION: get_fundamental_analysis(NVDA)
+After tool result: FINAL: <full scorecard output from tool>
+ 
+User: Give me a fundamental valuation of Grupo Bimbo.
+ACTION: get_fundamental_analysis(BIMBOA.MX)
+After tool result: FINAL: <full scorecard output from tool>
+ 
+User: Is Tesla a good investment right now?
+ACTION: get_fundamental_analysis(TSLA)
+After tool result: FINAL: <full scorecard output from tool>
+
+User: What is the 28-day CETES rate?
+ACTION: get_cetes_rate(28)
+After tool result: FINAL: The CETES 28-day rate (CETES28D) is 9.0000% as of 20/03/2025.
+
+User: What is the current tiee rate for 91 days as of 20/03/2025?
+ACTION: get_tiie_rate(91, 2025-03-20)
+After tool result: FINAL: The TIIE 91-day rate (TIIE91D) is 9.5000% as of 20/03/2025.
 
 User: What is the current Banxico target rate?
 ACTION: get_target_interest_rate_mexico()
@@ -272,10 +309,6 @@ User: What was the monthly inflation in Mexico in mid-2023?
 ACTION: get_mensual_inflation_mexico(2023-06-15)
 After tool result: FINAL: The monthly inflation rate in Mexico nearest to June 15, 2023 was 0.2200% as of 15/06/2023.
 
-User: What is the TIE 182-day rate?
-ACTION: get_tie_182()
-After tool result: FINAL: The TIE 182-day rate (most recent) is 9.4500% as of 15/04/2025.
-
 User: What is the current UDI value?
 ACTION: get_udis()
 After tool result: FINAL: The value of UDIs in Mexico (most recent) is 8.2341 MXN as of 30/04/2025.
@@ -298,4 +331,4 @@ ACTION: get_price_on_date(AAPL, March 2023)          <- date must be in YYYY-MM-
 ACTION: min_variance_portfolio("AAPL, MSFT, GOOGL")  <- never pack tickers into one string argument
 FINAL: Apple's price is around $210                  <- invented value, no tool was called
 FINAL: The CETES rate is roughly 9%                  <- recalled from memory, no tool was called
-"""
+"""

tools.py

@@ -13,7 +13,9 @@ import torch
 import torch.nn.functional as F
 
 import os
+from dotenv import load_dotenv
 
+load_dotenv()
 BANXICO_TOKEN = os.getenv("BANXICO_TOKEN")
 HF_LOGIN_KEY = os.getenv("HF_LOGIN_KEY")
 if HF_LOGIN_KEY:
@@ -59,25 +61,25 @@ def build_default_tool_registry() -> ToolRegistry:
 @tool("get_price_on_date")
 def get_price_on_date(ticker, date=None):
     t = yf.Ticker(ticker)
-
+    
     use_default_date = date is None
-
+    
     if date is None:
         date = datetime.date.today()
     else:
         date = datetime.datetime.strptime(date, "%Y-%m-%d").date()
-
-    data = pd.DataFrame(t.history(start=date - datetime.timedelta(days=5), end=date + datetime.timedelta(days=5))['Close'])
+    
+    data = pd.DataFrame(t.history(start=date - datetime.timedelta(days=5), end=date + datetime.timedelta(days=5))['Close'])    
     if data.empty:
         return f"No price data available for {t.ticker} around {date}."
-
+    
     data['Date'] = data.index.date
     data['DateDiff'] = np.abs(data['Date'] - date)
     nearest_row = data.loc[data['DateDiff'].idxmin()]
     price = nearest_row['Close']
     actual_date = nearest_row['Date']
     official_name = t.info['longName']
-
+    
     if use_default_date:
         return f"The last price of {official_name} ({t.ticker}) is ${price:.2f} as of {actual_date}."
     else:
@@ -160,115 +162,31 @@ def min_target_semivariance_portfolio(*tickers: str) -> str:
         f"Expected annual return: {(returns.mean() @ result.x * 252):.2%}\n"
         f"Annualized volatility: {(np.sqrt(result.x.T @ cov_matrix @ result.x) * np.sqrt(252)):.2%}"
     )
+    
+CETES_SERIES = {
+    28:  "SF43936",
+    91:  "SF43939",
+    182: "SF43942",
+    364: "SF43945",
+    728: "SF349785",
+}
 
-@tool("get_cetes_28")
-def get_cetes_28(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43936/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
-
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The CETES 28-day rate ({label}) is {value:.4f}% as of {fecha}."
-
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching CETES 28-day rate: {exc}"
-
-@tool("get_cetes_91")
-def get_cetes_91(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43939/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
-
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The CETES 91-day rate ({label}) is {value:.4f}% as of {fecha}."
+@tool("get_cetes_rate")
+def get_cetes_rate(term_days: int, date: str | None = None) -> str:
+    #valid days are the ones displayed above
+    if term_days not in CETES_SERIES:
+        valid = ", ".join(str(k) for k in CETES_SERIES)
+        return f"Invalid term '{term_days}'. Valid options are: {valid}."
 
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching CETES 91-day rate: {exc}"
-
-@tool("get_cetes_182")
-def get_cetes_182(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43942/datos"
+    series_id = CETES_SERIES[term_days]
+    url = f"https://www.banxico.org.mx/SieAPIRest/service/v1/series/{series_id}/datos"
     headers = {
         "Bmx-Token": BANXICO_TOKEN,
         "Content-Type": "application/json",
     }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
 
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The CETES 182-day rate ({label}) is {value:.4f}% as of {fecha}."
-
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching CETES 182-day rate: {exc}"
-
-@tool("get_cetes_364")
-def get_cetes_364(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43945/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
     try:
-        response = requests.get(URL, headers=headers)
+        response = requests.get(url, headers=headers)
         response.raise_for_status()
 
         obs_list = response.json()["bmx"]["series"][0]["datos"]
@@ -282,49 +200,16 @@ def get_cetes_364(date: str | None = None) -> str:
                 key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
             )
 
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
+        fecha = datetime.datetime.strptime(obs["fecha"], "%d/%m/%Y").strftime("%Y-%m-%d")
         value = float(obs["dato"])
         label = f"nearest to {date}" if date else "most recent"
-        return f"The CETES 364-day rate ({label}) is {value:.4f}% as of {fecha}."
+        return f"The CETES {term_days}-day rate ({label}) is {value:.4f}% as of {fecha}."
 
     except ValueError:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
-        return f"Error fetching CETES 364-day rate: {exc}"
-
-@tool("get_cetes_728")
-def get_cetes_728(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF349785/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
-
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The CETES 728-day rate ({label}) is {value:.4f}% as of {fecha}."
+        return f"Error fetching CETES {term_days}-day rate: {exc}"
 
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching CETES 728-day rate: {exc}"
 
 @tool("get_mensual_inflation_mexico")
 def get_mensual_inflation_mexico(date: str | None = None) -> str:
@@ -353,12 +238,12 @@ def get_mensual_inflation_mexico(date: str | None = None) -> str:
         value = float(obs["dato"])
         label = f"nearest to {date}" if date else "most recent"
         return f"The monthly inflation rate in Mexico ({label}) is {value:.4f}% as of {fecha}."
-
+    
     except ValueError:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
         return f"Error fetching monthly inflation rate in Mexico: {exc}"
-
+    
 @tool("get_inflation_mexico")
 def get_inflation_mexico(date: str | None = None) -> str:
     URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SP30577/datos"
@@ -391,7 +276,7 @@ def get_inflation_mexico(date: str | None = None) -> str:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
         return f"Error fetching annual inflation rate in Mexico: {exc}"
-
+    
 @tool("get_udis")
 def get_udis(date: str | None = None) -> str:
     URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SP68257/datos"
@@ -424,82 +309,33 @@ def get_udis(date: str | None = None) -> str:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
         return f"Error fetching UDIs value in Mexico: {exc}"
+    
+TIIE_SERIES = {
+    28:  "SF43783",
+    91:  "SF43878",
+    182: "SF111916",
+}
 
-@tool("get_tie_28")
-def get_tie_28(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43783/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
-
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The TIE 28-day rate ({label}) is {value:.4f}% as of {fecha}."
-
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching TIE 28-day rate: {exc}"
-
-@tool("get_tie_91")
-def get_tie_91(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF43878/datos"
+@tool("get_tiie_rate")
+def get_tiie_rate(term_days: int, date: str | None = None) -> str:
+    """
+    Fetches the TIIE (Tasa de Interés Interbancaria de Equilibrio) rate
+    for a given term in days from Banxico.
+    Valid terms are: 28, 91, 182.
+    """
+    if term_days not in TIIE_SERIES:
+        valid = ", ".join(str(k) for k in TIIE_SERIES)
+        return f"Invalid term '{term_days}'. Valid options are: {valid}."
+
+    series_id = TIIE_SERIES[term_days]
+    url = f"https://www.banxico.org.mx/SieAPIRest/service/v1/series/{series_id}/datos"
     headers = {
         "Bmx-Token": BANXICO_TOKEN,
         "Content-Type": "application/json",
     }
-    try:
-        response = requests.get(URL, headers=headers)
-        response.raise_for_status()
 
-        obs_list = response.json()["bmx"]["series"][0]["datos"]
-
-        if date is None:
-            obs = obs_list[-1]
-        else:
-            target = datetime.datetime.strptime(date, "%Y-%m-%d")
-            obs = min(
-                obs_list,
-                key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
-            )
-
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
-        value = float(obs["dato"])
-        label = f"nearest to {date}" if date else "most recent"
-        return f"The TIE 91-day rate ({label}) is {value:.4f}% as of {fecha}."
-
-    except ValueError:
-        return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
-    except Exception as exc:
-        return f"Error fetching TIE 91-day rate: {exc}"
-
-@tool("get_tie_182")
-def get_tie_182(date: str | None = None) -> str:
-    URL = "https://www.banxico.org.mx/SieAPIRest/service/v1/series/SF111916/datos"
-    headers = {
-        "Bmx-Token": BANXICO_TOKEN,
-        "Content-Type": "application/json",
-    }
     try:
-        response = requests.get(URL, headers=headers)
+        response = requests.get(url, headers=headers)
         response.raise_for_status()
 
         obs_list = response.json()["bmx"]["series"][0]["datos"]
@@ -513,16 +349,15 @@ def get_tie_182(date: str | None = None) -> str:
                 key=lambda o: abs(datetime.datetime.strptime(o["fecha"], "%d/%m/%Y") - target),
             )
 
-        fecha = obs["fecha"]
-        fecha = datetime.datetime.strptime(fecha, "%d/%m/%Y").strftime("%Y-%m-%d")
+        fecha = datetime.datetime.strptime(obs["fecha"], "%d/%m/%Y").strftime("%Y-%m-%d")
         value = float(obs["dato"])
         label = f"nearest to {date}" if date else "most recent"
-        return f"The TIE 182-day rate ({label}) is {value:.4f}% as of {fecha}."
+        return f"The TIIE {term_days}-day rate ({label}) is {value:.4f}% as of {fecha}."
 
     except ValueError:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
-        return f"Error fetching TIE 182-day rate: {exc}"
+        return f"Error fetching TIIE {term_days}-day rate: {exc}"
 
 @tool("get_target_interest_rate_mexico")
 def get_target_interest_rate_mexico(date: str | None = None) -> str:
@@ -556,44 +391,44 @@ def get_target_interest_rate_mexico(date: str | None = None) -> str:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
         return f"Error fetching target interest rate in Mexico: {exc}"
-
+    
 @tool("get_exchange_rate")
 def get_exchange_rate(base: str, quote: str, date: str | None = None) -> str:
     base = base.strip().upper()
     quote = quote.strip().upper()
     ticker_symbol = f"{base}{quote}=X"
-
+ 
     try:
         if date is None:
             target_date = datetime.date.today()
         else:
             target_date = datetime.datetime.strptime(date, "%Y-%m-%d").date()
-
+ 
         t = yf.Ticker(ticker_symbol)
         data = t.history(
             start=target_date - datetime.timedelta(days=7),
             end=target_date + datetime.timedelta(days=7),
         )
-
+ 
         if data.empty:
             return (
                 f"No exchange rate data found for {base}/{quote} ({ticker_symbol}). "
                 f"Verify that both currency codes are valid ISO 4217 codes."
             )
-
+ 
         data["Date"] = data.index.date
         data["DateDiff"] = data["Date"].apply(lambda d: abs((d - target_date).days))
         nearest = data.loc[data["DateDiff"].idxmin()]
         rate = nearest["Close"]
         actual_date = nearest["Date"]
         date_label = f"nearest to {date}" if date else "most recent"
-        return f"The exchange rate for {base}/{quote} ({date_label}) is {rate:.6f} as of {actual_date}."
-
+        return f"The exchange rate for {base}/{quote} ({date_label}) is {rate:.6f} as of {actual_date}." 
+        
     except ValueError:
         return f"Invalid date format '{date}'. Please use YYYY-MM-DD (e.g. 2024-01-15)."
     except Exception as exc:
         return f"Error fetching exchange rate for {base}/{quote}: {exc}"
-
+    
 def _get_news(ticker: str) -> list[dict]:
     t = yf.Ticker(ticker)
     news = t.news
@@ -627,6 +462,11 @@ _LABEL_TO_SCORE = {
     "negative": -1
 }
 
+_SCORE_TO_LABEL = {
+    lambda s: s > 0.15: "positive",
+    lambda s: s < -0.15: "negative",    
+}
+
 def _bucket_label(score: float) -> str:
     if score > 0.15:
         return "positive"
@@ -635,7 +475,7 @@ def _bucket_label(score: float) -> str:
     return "neutral"
 
 def _recency_weights(pub_dates: list[str]) -> list[float]:
-    decay = 0.01
+    decay = 0.01  
     parsed = []
     for d in pub_dates:
         try:
@@ -659,6 +499,7 @@ def _recency_weights(pub_dates: list[str]) -> list[float]:
     return weights
 
 def _score_texts(texts: list[str]) -> list[dict]:
+    """Returns a list of {label, confidence} dicts, one per input text."""
     tokenizer, model = _load_finbert()
     results = []
     with torch.no_grad():
@@ -683,6 +524,8 @@ def _score_texts(texts: list[str]) -> list[dict]:
 
 @tool("get_news_sentiment")
 def get_news_sentiment(ticker: str) -> str:
+    """Fetches recent news for a ticker and returns a FinBERT-based
+    sentiment score aggregated across all available headlines."""
     articles = _get_news(ticker)
     comp_name = yf.Ticker(ticker).info.get("longName", ticker)
     if not articles:
@@ -729,6 +572,237 @@ def get_news_sentiment(ticker: str) -> str:
         f"Composite score: {composite:+.4f} ({label}). "
         f"Top influencing headlines: {top_headlines}"
     )
+    
+@tool("calculate_inflation_impact")
+def calculate_inflation_impact(amount: float, months: int, annual_inflation_rate: float) -> str:
+    monthly_rate = (1 + annual_inflation_rate / 100) ** (1 / 12) - 1
+    future_equivalent = amount * (1 + monthly_rate) ** months
+    purchasing_power_loss = future_equivalent - amount
+    effective_value = amount - purchasing_power_loss
+
+    return (
+        f"With an annual inflation rate of {annual_inflation_rate:.2f}%, "
+        f"{amount:.2f} pesos today will have the purchasing power of "
+        f"{effective_value:.2f} pesos after {months} month(s). "
+        f"That is a loss of {purchasing_power_loss:.2f} pesos in real value."
+    )
+
+@tool("multiply")
+def multiply(a: float, b: float) -> str:
+    result = a * b
+    return f"The result of {a} × {b} is {result}."
+
+# --------------- sector-adjusted valuation thresholds -------------------------
+# Tuple layout: (pe_strong, pe_weak, pb_strong, pb_weak, ev_ebitda_strong, ev_ebitda_weak)
+# "strong" means the value that earns the maximum score of 2.
+# "weak"   means the value that earns the minimum score of 0.
+# Values between the two thresholds score 1 (neutral).
+_SECTOR_VALUATION_THRESHOLDS: dict[str, tuple] = {
+    "technology":         (25, 45, 4.0, 10.0, 15, 30),
+    "healthcare":         (20, 35, 3.0,  8.0, 14, 25),
+    "financial-services": (12, 20, 1.0,  2.5, 10, 18),
+    "consumer-cyclical":  (18, 30, 2.5,  6.0, 12, 22),
+    "consumer-defensive": (18, 28, 3.0,  6.0, 12, 20),
+    "energy":             (10, 20, 1.5,  3.0,  6, 14),
+    "basic-materials":    (12, 22, 1.5,  3.5,  8, 16),
+    "industrials":        (18, 30, 2.5,  5.0, 12, 20),
+    "real-estate":        (30, 55, 1.5,  3.5, 18, 30),
+    "utilities":          (15, 25, 1.5,  3.0, 10, 18),
+    "default":            (18, 35, 2.5,  6.0, 12, 22),
+}
+ 
+ 
+def _valuation_thresholds(sector_key: str | None) -> tuple:
+    key = (sector_key or "").lower()
+    return _SECTOR_VALUATION_THRESHOLDS.get(key, _SECTOR_VALUATION_THRESHOLDS["default"])
+ 
+ 
+def _score_metric(value: float, strong_threshold: float, weak_threshold: float,
+                  lower_is_better: bool = True) -> int:
+    """
+    Scores a single metric on a 0–2 scale.
+ 
+    For lower_is_better metrics (P/E, D/E, EV/EBITDA …):
+        value <= strong_threshold → 2
+        value >= weak_threshold   → 0
+        in between                → 1
+ 
+    For higher_is_better metrics (ROE, margins, FCF yield …):
+        value >= strong_threshold → 2
+        value <= weak_threshold   → 0
+        in between                → 1
+    """
+    if lower_is_better:
+        if value <= strong_threshold:
+            return 2
+        if value >= weak_threshold:
+            return 0
+        return 1
+    else:
+        if value >= strong_threshold:
+            return 2
+        if value <= weak_threshold:
+            return 0
+        return 1
+ 
+ 
+@tool("get_fundamental_analysis")
+def get_fundamental_analysis(ticker: str) -> str:
+    """
+    Performs a quantitative fundamental analysis scorecard for a given ticker.
+ 
+    Evaluates 15 metrics across four categories:
+      - Valuation      (P/E, P/B, EV/EBITDA, PEG)               max  8 pts
+      - Profitability  (ROE, ROA, Gross margin, Net margin)       max  8 pts
+      - Financial Health (D/E, Current ratio, IC, FCF yield)     max  8 pts
+      - Growth         (Revenue growth, Earnings growth, Div yield) max 6 pts
+                                                                 ──────────
+                                                          TOTAL   max 30 pts
+ 
+    Scoring per metric: 2 = strong, 1 = neutral / data unavailable, 0 = weak.
+    Valuation thresholds are sector-adjusted via yfinance sectorKey.
+    Composite: ≥70% → BUY | 40–69% → HOLD | <40% → SELL.
+ 
+    Apply the same exchange suffix rules as get_price_on_date (e.g. BIMBOA.MX).
+    """
+    t    = yf.Ticker(ticker)
+    info = t.info
+ 
+    company_name = info.get("longName", ticker)
+    sector_key   = info.get("sectorKey", None)
+    sector_label = info.get("sector", "Unknown sector")
+ 
+    pe_s, pe_w, pb_s, pb_w, ev_s, ev_w = _valuation_thresholds(sector_key)
+ 
+    def safe(key: str, scale: float = 1.0):
+        """Returns (scaled_value, is_available). Missing or non-numeric → (None, False)."""
+        raw = info.get(key)
+        if raw is None or not isinstance(raw, (int, float)):
+            return None, False
+        return raw * scale, True
+ 
+    # ── Valuation (max 8 pts) ──────────────────────────────────────────────────
+    pe,        pe_ok  = safe("trailingPE")
+    pb,        pb_ok  = safe("priceToBook")
+    ev_ebitda, ev_ok  = safe("enterpriseToEbitda")
+    peg,       peg_ok = safe("pegRatio")
+ 
+    pe_score  = _score_metric(pe,        pe_s, pe_w, lower_is_better=True) if pe_ok  else 1
+    pb_score  = _score_metric(pb,        pb_s, pb_w, lower_is_better=True) if pb_ok  else 1
+    ev_score  = _score_metric(ev_ebitda, ev_s, ev_w, lower_is_better=True) if ev_ok  else 1
+    peg_score = _score_metric(peg,       1.0,  2.0,  lower_is_better=True) if peg_ok else 1
+ 
+    valuation_score = pe_score + pb_score + ev_score + peg_score  # max 8
+ 
+    # ── Profitability (max 8 pts) ──────────────────────────────────────────────
+    roe,     roe_ok = safe("returnOnEquity", scale=100)
+    roa,     roa_ok = safe("returnOnAssets", scale=100)
+    gross_m, gm_ok  = safe("grossMargins",   scale=100)
+    net_m,   nm_ok  = safe("profitMargins",  scale=100)
+ 
+    roe_score = _score_metric(roe,     15.0, 8.0,  lower_is_better=False) if roe_ok else 1
+    roa_score = _score_metric(roa,      5.0, 2.0,  lower_is_better=False) if roa_ok else 1
+    gm_score  = _score_metric(gross_m, 40.0, 20.0, lower_is_better=False) if gm_ok  else 1
+    nm_score  = _score_metric(net_m,   10.0,  3.0, lower_is_better=False) if nm_ok  else 1
+ 
+    profit_score = roe_score + roa_score + gm_score + nm_score  # max 8
+ 
+    # ── Financial Health (max 8 pts) ───────────────────────────────────────────
+    de,      de_ok   = safe("debtToEquity")
+    cr,      cr_ok   = safe("currentRatio")
+    ebitda,  ebit_ok = safe("ebitda")
+    int_exp, ie_ok   = safe("interestExpense")
+    fcf,     fcf_ok  = safe("freeCashflow")
+    mktcap,  mc_ok   = safe("marketCap")
+ 
+    # yfinance returns D/E as a percentage (e.g. 150 means 1.50); normalise to ratio.
+    de_adj   = de / 100.0 if de_ok else None
+    de_score = _score_metric(de_adj, 0.5, 1.5, lower_is_better=True) if de_ok else 1
+ 
+    cr_score = _score_metric(cr, 2.0, 1.0, lower_is_better=False) if cr_ok else 1
+ 
+    # Interest coverage = EBITDA / |interest expense|; higher is better.
+    if ebit_ok and ie_ok and int_exp != 0:
+        ic       = abs(ebitda) / abs(int_exp)
+        ic_score = _score_metric(ic, 5.0, 2.0, lower_is_better=False)
+    else:
+        ic       = None
+        ic_score = 1
+ 
+    # FCF yield = FCF / market cap (%); >5% strong, <0% weak.
+    if fcf_ok and mc_ok and mktcap > 0:
+        fcf_yield = (fcf / mktcap) * 100
+        fcf_score = _score_metric(fcf_yield, 5.0, 0.0, lower_is_better=False)
+    else:
+        fcf_yield = None
+        fcf_score = 1
+ 
+    health_score = de_score + cr_score + ic_score + fcf_score  # max 8
+ 
+    # ── Growth (max 6 pts) ─────────────────────────────────────────────────────
+    rev_g,  rg_ok = safe("revenueGrowth",  scale=100)
+    earn_g, eg_ok = safe("earningsGrowth", scale=100)
+    div_y,  dy_ok = safe("dividendYield",  scale=100)
+ 
+    rev_score  = _score_metric(rev_g,  10.0, 0.0, lower_is_better=False) if rg_ok else 1
+    earn_score = _score_metric(earn_g, 10.0, 0.0, lower_is_better=False) if eg_ok else 1
+ 
+    # Dividend yield: 2–5.5% is the ideal income range.
+    # Below 0.5% is neutral (growth company, no penalty). Above 6% may signal distress.
+    if dy_ok:
+        if 2.0 <= div_y <= 5.5:
+            div_score = 2
+        elif div_y > 6.0 or div_y < 0.5:
+            div_score = 0
+        else:
+            div_score = 1
+    else:
+        div_score = 1  # no dividend data → neutral
+ 
+    growth_score = rev_score + earn_score + div_score  # max 6
+ 
+    # ── Composite & recommendation ─────────────────────────────────────────────
+    MAX_SCORE = 30
+    composite = valuation_score + profit_score + health_score + growth_score
+    pct       = composite / MAX_SCORE
+ 
+    if pct >= 0.70:
+        recommendation = "BUY"
+        rationale      = "the company scores strongly across most fundamental dimensions"
+    elif pct >= 0.40:
+        recommendation = "HOLD"
+        rationale      = "the fundamentals are mixed with no compelling entry or exit signal"
+    else:
+        recommendation = "SELL"
+        rationale      = "the company shows material weakness across multiple fundamental dimensions"
+ 
+    def fmt(value, decimals: int = 2, suffix: str = "") -> str:
+        return "N/A" if value is None else f"{value:.{decimals}f}{suffix}"
+ 
+    return (
+        f"Fundamental analysis scorecard for {company_name} ({ticker.upper()}) | Sector: {sector_label}. "
+        f"VALUATION ({valuation_score}/8): "
+        f"P/E {fmt(pe)}x [score {pe_score}/2], "
+        f"P/B {fmt(pb)}x [score {pb_score}/2], "
+        f"EV/EBITDA {fmt(ev_ebitda)}x [score {ev_score}/2], "
+        f"PEG {fmt(peg)} [score {peg_score}/2]. "
+        f"PROFITABILITY ({profit_score}/8): "
+        f"ROE {fmt(roe)}% [score {roe_score}/2], "
+        f"ROA {fmt(roa)}% [score {roa_score}/2], "
+        f"Gross margin {fmt(gross_m)}% [score {gm_score}/2], "
+        f"Net margin {fmt(net_m)}% [score {nm_score}/2]. "
+        f"FINANCIAL HEALTH ({health_score}/8): "
+        f"D/E {fmt(de_adj)} [score {de_score}/2], "
+        f"Current ratio {fmt(cr)} [score {cr_score}/2], "
+        f"Interest coverage {fmt(ic)}x [score {ic_score}/2], "
+        f"FCF yield {fmt(fcf_yield)}% [score {fcf_score}/2]. "
+        f"GROWTH ({growth_score}/6): "
+        f"Revenue growth {fmt(rev_g)}% [score {rev_score}/2], "
+        f"Earnings growth {fmt(earn_g)}% [score {earn_score}/2], "
+        f"Dividend yield {fmt(div_y)}% [score {div_score}/2]. "
+        f"COMPOSITE SCORE: {composite}/{MAX_SCORE} ({pct:.0%}). "
+        f"RECOMMENDATION: {recommendation} — {rationale}."
+    )
 
 @tool("respond_to_greeting")
 def respond_to_greeting() -> str:
@@ -737,3 +811,4 @@ def respond_to_greeting() -> str:
 @tool("respond_no_available_tool")
 def respond_no_available_tool(tool_name: str) -> str:
     return f"Sorry, currently i'm capable of doing that. Check the list of avaiable tools for more information."
+