Hotfix (#55)

* fixed issue with tools using agent_config properly (when specified)

* upgrade openai llm to support o3-mini

* updated

* updated README

* fixed README

* fixes

.pylintrc

@@ -22,4 +22,5 @@ disable =
     too-many-branches,
     too-many-instance-attributes,
     too-many-arguments,
+    too-many-positional-arguments,
 

README.md

@@ -198,11 +198,22 @@ similar queries that require a response in terms of a list of matching documents
 
 ## 🛠️ Agent Tools at a Glance
 
-`vectara-agentic` provides a few tools out of the box:
+`vectara-agentic` provides a few tools out of the box (see ToolsCatalog for details):
+
 1. **Standard tools**: 
 - `summarize_text`: a tool to summarize a long text into a shorter summary (uses LLM)
 - `rephrase_text`: a tool to rephrase a given text, given a set of rephrase instructions (uses LLM)
-
+These tools use an LLM and so would use the `Tools` LLM specified in your `AgentConfig`.
+To instantiate them:
+
+```python
+from vectara_agentic.tools_catalog import ToolsCatalog
+summarize_text = ToolsCatalog(agent_config).summarize_text
+```
+
+This ensures the summarize_text tool is configured with the proper LLM provider and model as 
+specified in the Agent configuration.
+
 2. **Legal tools**: a set of tools for the legal vertical, such as:
 - `summarize_legal_text`: summarize legal text with a certain point of view
 - `critique_as_judge`: critique a legal text as a judge, providing their perspective
@@ -239,19 +250,44 @@ mult_tool = ToolsFactory().create_tool(mult_func)
 
 ## 🛠️ Configuration
 
+## Configuring Vectara-agentic
+
 The main way to control the behavior of `vectara-agentic` is by passing an `AgentConfig` object to your `Agent` when creating it.
-This object will include the following items:
-- `VECTARA_AGENTIC_AGENT_TYPE`: valid values are `REACT`, `LLMCOMPILER`, `LATS` or `OPENAI` (default: `OPENAI`)
-- `VECTARA_AGENTIC_MAIN_LLM_PROVIDER`: valid values are `OPENAI`, `ANTHROPIC`, `TOGETHER`, `GROQ`, `COHERE`, `BEDROCK`, `GEMINI` or `FIREWORKS` (default: `OPENAI`)
-- `VECTARA_AGENTIC_MAIN_MODEL_NAME`: agent model name (default depends on provider)
-- `VECTARA_AGENTIC_TOOL_LLM_PROVIDER`: tool LLM provider (default: `OPENAI`)
-- `VECTARA_AGENTIC_TOOL_MODEL_NAME`: tool model name (default depends on provider)
-- `VECTARA_AGENTIC_OBSERVER_TYPE`: valid values are `ARIZE_PHOENIX` or `NONE` (default: `NONE`)
-- `VECTARA_AGENTIC_API_KEY`: a secret key if using the API endpoint option (defaults to `dev-api-key`)
+For example:
+
+```python
+agent_config = AgentConfig(
+    agent_type = AgentType.REACT,
+    main_llm_provider = ModelProvider.ANTHROPIC,
+    main_llm_model_name = 'claude-3-5-sonnet-20241022',
+    tool_llm_provider = ModelProvider.TOGETHER,
+    tool_llm_model_name = 'meta-llama/Llama-3.3-70B-Instruct-Turbo'
+)
+
+agent = Agent(
+    tools=[query_financial_reports_tool],
+    topic="10-K financial reports",
+    custom_instructions="You are a helpful financial assistant in conversation with a user.",
+    agent_config=agent_config
+)
+```
+
+The `AgentConfig` object may include the following items:
+- `agent_type`: the agent type. Valid values are `REACT`, `LLMCOMPILER`, `LATS` or `OPENAI` (default: `OPENAI`).
+- `main_llm_provider` and `tool_llm_provider`: the LLM provider for main agent and for the tools. Valid values are `OPENAI`, `ANTHROPIC`, `TOGETHER`, `GROQ`, `COHERE`, `BEDROCK`, `GEMINI` or `FIREWORKS` (default: `OPENAI`).
+- `main_llm_model_name` and `tool_llm_model_name`: agent model name for agent and tools (default depends on provider).
+- `observer`: the observer type; should be `ARIZE_PHOENIX` or if undefined no observation framework will be used.
+- `endpoint_api_key`: a secret key if using the API endpoint option (defaults to `dev-api-key`)
 
 If any of these are not provided, `AgentConfig` first tries to read the values from the OS environment.
 
-When creating a `VectaraToolFactory`, you can pass in a `vectara_api_key`, `vectara_customer_id`, and `vectara_corpus_id` to the factory. If not passed in, it will be taken from the environment variables (`VECTARA_API_KEY`, `VECTARA_CUSTOMER_ID` and `VECTARA_CORPUS_ID`). Note that `VECTARA_CORPUS_ID` can be a single ID or a comma-separated list of IDs (if you want to query multiple corpora).
+## Configuring Vectara RAG or search tools
+
+When creating a `VectaraToolFactory`, you can pass in a `vectara_api_key`, `vectara_customer_id`, and `vectara_corpus_id` to the factory. 
+
+If not passed in, it will be taken from the environment variables (`VECTARA_API_KEY`, `VECTARA_CUSTOMER_ID` and `VECTARA_CORPUS_ID`). Note that `VECTARA_CORPUS_ID` can be a single ID or a comma-separated list of IDs (if you want to query multiple corpora).
+
+These values will be used as credentials when creating Vectara tools - in `create_rag_tool()` and `create_search_tool()`.
 
 ## ℹ️ Additional Information
 

requirements.txt

@@ -3,7 +3,7 @@ llama-index-indices-managed-vectara==0.3.1
 llama-index-agent-llm-compiler==0.3.0
 llama-index-agent-lats==0.3.0
 llama-index-agent-openai==0.4.3
-llama-index-llms-openai==0.3.16
+llama-index-llms-openai==0.3.18
 llama-index-llms-anthropic==0.6.4
 llama-index-llms-together==0.3.1
 llama-index-llms-groq==0.3.1

vectara_agentic/_version.py

@@ -1,4 +1,4 @@
 """
 Define the version of the package.
 """
-__version__ = "0.1.25"
+__version__ = "0.1.26"

vectara_agentic/agent.py

@@ -397,8 +397,8 @@ def report(self) -> None:
                 print(f"- {tool.metadata.name}")
             else:
                 print("- tool without metadata")
-        print(f"Agent LLM = {get_llm(LLMRole.MAIN).metadata.model_name}")
-        print(f"Tool LLM = {get_llm(LLMRole.TOOL).metadata.model_name}")
+        print(f"Agent LLM = {get_llm(LLMRole.MAIN, config=self.agent_config).metadata.model_name}")
+        print(f"Tool LLM = {get_llm(LLMRole.TOOL, config=self.agent_config).metadata.model_name}")
 
     def token_counts(self) -> dict:
         """

vectara_agentic/tools.py

@@ -19,9 +19,10 @@
 
 
 from .types import ToolType
-from .tools_catalog import summarize_text, rephrase_text, critique_text, get_bad_topics
+from .tools_catalog import ToolsCatalog, get_bad_topics
 from .db_tools import DBLoadSampleData, DBLoadUniqueValues, DBLoadData
 from .utils import is_float
+from .agent_config import AgentConfig
 
 LI_packages = {
     "yahoo_finance": ToolType.QUERY,
@@ -624,6 +625,9 @@ class ToolsFactory:
     A factory class for creating agent tools.
     """
 
+    def __init__(self, agent_config: AgentConfig = None) -> None:
+        self.agent_config = agent_config
+
     def create_tool(self, function: Callable, tool_type: ToolType = ToolType.QUERY) -> VectaraTool:
         """
         Create a tool from a function.
@@ -686,7 +690,8 @@ def standard_tools(self) -> List[FunctionTool]:
         """
         Create a list of standard tools.
         """
-        return [self.create_tool(tool) for tool in [summarize_text, rephrase_text]]
+        tc = ToolsCatalog(self.agent_config)
+        return [self.create_tool(tool) for tool in [tc.summarize_text, tc.rephrase_text, tc.critique_text]]
 
     def guardrail_tools(self) -> List[FunctionTool]:
         """
@@ -711,15 +716,17 @@ def summarize_legal_text(
             """
             Use this tool to summarize legal text with no more than summary_max_length characters.
             """
-            return summarize_text(text, expertise="law")
+            tc = ToolsCatalog(self.agent_config)
+            return tc.summarize_text(text, expertise="law")
 
         def critique_as_judge(
             text: str = Field(description="the original text."),
         ) -> str:
             """
             Critique the legal document.
             """
-            return critique_text(
+            tc = ToolsCatalog(self.agent_config)
+            return tc.critique_text(
                 text,
                 role="judge",
                 point_of_view="""

vectara_agentic/tools_catalog.py

@@ -2,13 +2,15 @@
 This module contains the tools catalog for the Vectara Agentic.
 """
 from typing import List
-from functools import lru_cache
 from datetime import date
+
+from inspect import signature
 import requests
 
 from pydantic import Field
 
 from .types import LLMRole
+from .agent_config import AgentConfig
 from .utils import get_llm
 
 req_session = requests.Session()
@@ -27,97 +29,114 @@ def get_current_date() -> str:
     """
     return date.today().strftime("%A, %B %d, %Y")
 
-#
-# Standard Tools
-#
-@lru_cache(maxsize=None)
-def summarize_text(
-    text: str = Field(description="the original text."),
-    expertise: str = Field(
-        description="the expertise to apply to the summarization.",
-    ),
-) -> str:
-    """
-    This is a helper tool.
-    Use this tool to summarize text using a given expertise
-    with no more than summary_max_length characters.
-
-    Args:
-        text (str): The original text.
-        expertise (str): The expertise to apply to the summarization.
-
-    Returns:
-        str: The summarized text.
-    """
-    if not isinstance(expertise, str):
-        return "Please provide a valid string for expertise."
-    if not isinstance(text, str):
-        return "Please provide a valid string for text."
-    expertise = "general" if len(expertise) < 3 else expertise.lower()
-    prompt = f"As an expert in {expertise}, summarize the provided text"
-    prompt += " into a concise summary."
-    prompt += f"\noriginal text: {text}\nsummary:"
-    llm = get_llm(LLMRole.TOOL)
-    response = llm.complete(prompt)
-    return response.text
-
-
-@lru_cache(maxsize=None)
-def rephrase_text(
-    text: str = Field(description="the original text."),
-    instructions: str = Field(description="the specific instructions for how to rephrase the text."),
-) -> str:
-    """
-    This is a helper tool.
-    Use this tool to rephrase the text according to the provided instructions.
-    For example, instructions could be "as a 5 year old would say it."
 
-    Args:
-        text (str): The original text.
-        instructions (str): The specific instructions for how to rephrase the text.
+def remove_self_from_signature(func):
+    """Decorator to remove 'self' from a method's signature for introspection."""
+    sig = signature(func)
+    params = list(sig.parameters.values())
+    # Remove the first parameter if it is named 'self'
+    if params and params[0].name == "self":
+        params = params[1:]
+    new_sig = sig.replace(parameters=params)
+    func.__signature__ = new_sig
+    return func
 
-    Returns:
-        str: The rephrased text.
+class ToolsCatalog:
     """
-    prompt = f"""
-    Rephrase the provided text according to the following instructions: {instructions}.
-    If the input is Markdown, keep the output in Markdown as well.
-    original text: {text}
-    rephrased text:
+    A curated set of tools for vectara-agentic
     """
-    llm = get_llm(LLMRole.TOOL)
-    response = llm.complete(prompt)
-    return response.text
-
-
-@lru_cache(maxsize=None)
-def critique_text(
-    text: str = Field(description="the original text."),
-    role: str = Field(default=None, description="the role of the person providing critique."),
-    point_of_view: str = Field(default=None, description="the point of view with which to provide critique."),
-) -> str:
-    """
-    This is a helper tool.
-    Critique the text from the specified point of view.
-
-    Args:
-        text (str): The original text.
-        role (str): The role of the person providing critique.
-        point_of_view (str): The point of view with which to provide critique.
-
-    Returns:
-        str: The critique of the text.
-    """
-    if role:
-        prompt = f"As a {role}, critique the provided text from the point of view of {point_of_view}."
-    else:
-        prompt = f"Critique the provided text from the point of view of {point_of_view}."
-    prompt += "Structure the critique as bullet points.\n"
-    prompt += f"Original text: {text}\nCritique:"
-    llm = get_llm(LLMRole.TOOL)
-    response = llm.complete(prompt)
-    return response.text
-
+    def __init__(self, agent_config: AgentConfig):
+        self.agent_config = agent_config
+
+    @remove_self_from_signature
+    def summarize_text(
+        self,
+        text: str = Field(description="the original text."),
+        expertise: str = Field(
+            description="the expertise to apply to the summarization.",
+        ),
+    ) -> str:
+        """
+        This is a helper tool.
+        Use this tool to summarize text using a given expertise
+        with no more than summary_max_length characters.
+
+        Args:
+            text (str): The original text.
+            expertise (str): The expertise to apply to the summarization.
+
+        Returns:
+            str: The summarized text.
+        """
+        if not isinstance(expertise, str):
+            return "Please provide a valid string for expertise."
+        if not isinstance(text, str):
+            return "Please provide a valid string for text."
+        expertise = "general" if len(expertise) < 3 else expertise.lower()
+        prompt = (
+            f"As an expert in {expertise}, summarize the provided text "
+            "into a concise summary.\n"
+            f"Original text: {text}\nSummary:"
+        )
+        llm = get_llm(LLMRole.TOOL, config=self.agent_config)
+        response = llm.complete(prompt)
+        return response.text
+
+    @remove_self_from_signature
+    def rephrase_text(
+        self,
+        text: str = Field(description="the original text."),
+        instructions: str = Field(description="the specific instructions for how to rephrase the text."),
+    ) -> str:
+        """
+        This is a helper tool.
+        Use this tool to rephrase the text according to the provided instructions.
+        For example, instructions could be "as a 5 year old would say it."
+
+        Args:
+            text (str): The original text.
+            instructions (str): The specific instructions for how to rephrase the text.
+
+        Returns:
+            str: The rephrased text.
+        """
+        prompt = (
+            f"Rephrase the provided text according to the following instructions: {instructions}.\n"
+            "If the input is Markdown, keep the output in Markdown as well.\n"
+            f"Original text: {text}\nRephrased text:"
+        )
+        llm = get_llm(LLMRole.TOOL, config=self.agent_config)
+        response = llm.complete(prompt)
+        return response.text
+
+    @remove_self_from_signature
+    def critique_text(
+        self,
+        text: str = Field(description="the original text."),
+        role: str = Field(default=None, description="the role of the person providing critique."),
+        point_of_view: str = Field(default=None, description="the point of view with which to provide critique."),
+    ) -> str:
+        """
+        This is a helper tool.
+        Critique the text from the specified point of view.
+
+        Args:
+            text (str): The original text.
+            role (str): The role of the person providing critique.
+            point_of_view (str): The point of view with which to provide critique.
+
+        Returns:
+            str: The critique of the text.
+        """
+        if role:
+            prompt = f"As a {role}, critique the provided text from the point of view of {point_of_view}."
+        else:
+            prompt = f"Critique the provided text from the point of view of {point_of_view}."
+        prompt += "\nStructure the critique as bullet points.\n"
+        prompt += f"Original text: {text}\nCritique:"
+        llm = get_llm(LLMRole.TOOL, config=self.agent_config)
+        response = llm.complete(prompt)
+        return response.text
 
 #
 # Guardrails tool: returns list of topics to avoid