chore: input prompts

task_1/input_prompt.py

@@ -0,0 +1,6 @@
+PROMPT = """
+Create a LangGraph-based text-to-SQL agent. It should generate sql, execute that sql against the chinook sqlite database, and then generate a response answer in natural language based on the result. 
+To use chinook db, fetch it from `https://raw.githubusercontent.com/lerocha/chinook-database/master/ChinookDatabase/DataSources/Chinook_Sqlite.sql` and create  in-memory db using langgraph. 
+It might be useful to pass in the detailed info of the table schema to the prompt so the agent can convert user's natural request to correct sql. Whenever, the query is irrelevant, or cannot be answered using the sql db search, just say you don't know the answer and don't talk about anything. 
+Your purpose is to only convert text request to sql and generate response in natural language.
+"""

task_2/input_prompt.py

@@ -0,0 +1,20 @@
+PROMPT = """
+Create me a company researcher built using langgraph. It should be a multi-node graph. The user should be expected to provide the name of the company and the optional notes if they want. There should be a set maximum search queries that we should do per company and max search results. The LLM should generate the queries that should be searched using the Tavily API to fill the following structured object
+
+  "title": "CompanyInfo",
+  "description": "Basic information about a company",
+  "type": "object",
+  "properties": {
+    "company_name": {"type": "string", "description": "Official name of the company"},
+    "founding_year": {"type": "integer", "description": "Year the company was founded"},
+    "founder_names": {"type": "array", "items": {"type": "string"}, "description": "Names of the founding team members"},
+    "product_description": {"type": "string", "description": "Brief description of the company's main product or service"},
+    "funding_summary": {"type": "string", "description": "Summary of the company's funding history”}
+    "notable_customers": {"type": "string", "description": "Known customers that use company's product/service"}
+
+  },
+  "required": ["company_name"]
+}
+
+There should be a reflection step at the end. This step should determine if we have good and sufficient information for the the company. If we don’t have sufficient information we should execute web searches again to get information on the company. There should also be a cap on the number of allowed reflection steps. Keep track of the conversation in messages array and try to do web searchers in parallel to improve speed.
+"""

task_3/input_prompt.py

@@ -0,0 +1,6 @@
+PROMPT = """
+You have been provided code for a memory agent. I want to extend its functionality. Right now, it stores memories user id wise. I want it to store user id and category wise. 
+While storing a memory it should determine if a memory is personal, professional, or other. It should then save based on that. While retrieving memories it should determine what category the message(s) are from. 
+Based on that, it should only retrieve relevant memories. Additionaly, there should be an interrupt before saving a memory. If the user inputs accept, the memory should get saved.
+If the user inputs anything else, the memory should be rejected. Fix any other bugs/issues as well.
+"""