ai-llm-agentic-tooling-langchain-langgraph

# AI LLM Agentic Tooling with LangChain/LangGraph Integrates LangChain and LangGraph to facilitate the development of LLM-powered agents and applications. This skill focuses on building advanced logic and workflows using these frameworks in Python. ## Use Cases Use this skill when: - Creating complex workflows involving decision-making models. - Building agents that require chaining multiple LLM calls efficiently. - Integrating external tools or APIs within a conversational agent framework. ## Implementation Patterns This skill offers an integration guide for LangChain and LangGraph, enabling the development of LLM-powered agents in Python. It facilitates various advanced logic and workflows through examples and best practices. ### Basic LangChain Agent Creation This example demonstrates how to define a basic LangChain agent: ```python from langchain import LLMChain, PromptTemplate # Define a simple LLM chain prompt = PromptTemplate(input_variables=['input'], template="""You are a helpful assistant. Assist with: {input}. """) agent = LLMChain(prompt=prompt) ``` ### LangGraph Workflow Example This section illustrates how to use LangGraph to define tasks: ```python from langgraph import Executor, Task @Task def fetch_data(): return {'data_key': 'value'} @Task def process_data(data): return data['data_key'] + ' processed' executor = Executor(tasks=[fetch_data, process_data]) executor.run() ``` ### Advanced Usage with Error Handling Include retries and error logging in your agents. Use state management techniques to preserve data between tasks. ## Constraints on Use - Ensure prompt structures are maintained to maximize performance and clarity. - Validate context objects to ensure they adhere to expected formats and types. ## Metadata Updates ```yaml archetypes: tactical anti_triggers: - vague conversation - overly generic request response_profile: verbosity: medium directive_strength: high abstraction_level: operational ``` ### Basic LangChain Agent Creation ```python from langchain import LLMChain, PromptTemplate # Define a simple LLM chain prompt = PromptTemplate(input_variables=['input'], template="""You are a helpful assistant. Assist with: {input}. """) agent = LLMChain(prompt=prompt) ``` ### LangGraph Workflow Example ```python from langgraph import Executor, Task # Define tasks and executor @Task def fetch_data(): return {'data_key': 'value'} @Task def process_data(data): return data['data_key'] + ' processed' executor = Executor(tasks=[fetch_data, process_data]) executor.run() ``` ### Advanced Usage with Error Handling - Include retries and error logging in your agents. - Use state management techniques to preserve data between tasks.