customized prompts

app/context_generator.py

@@ -58,7 +58,7 @@ For each topic, provide a separate context with the following structure:
 1. A concise introduction to the topic
 2. Key technical concepts and principles
 3. if the difficulty is hard give context paragraph from the very detail of the topic which even a 15+ years of experienced shouldn't know about. For easy 1-2 years of experience should be able to answer medium should be somewhere in between. 
-4.scenario specific texts eg: To encrypt a large dataset, symmetric encryption is preferred because of its superior speed compared to the computational overhead of asymmetric encryption.
+4.scenario specific texts eg: A developer is tasked with designing an API for a new social media platform. Considering factors like scalability and efficient data fetching, GraphQL would be a suitable choice.
 5. Important considerations or limitations.
 6. make sure the context will be used to genearte mcqs in future so make sure to include context that would yield mcqs that are harder to answer and make the users think thrice before answering a question.
 7. Don't include any special chars like ,/ etc that wouldn't be parsed by python's basic json parser as it would cause an erorr in the system
@@ -114,12 +114,11 @@ The JSON must be valid and properly formatted. Each topic's context should be 3-
 Generate a detailed and comprehensive context about the technical topic: {topics[0]}.
 The difficulty level for the context should be '{difficulty}'.
 Since this is a single focused topic, provide 3-4 detailed paragraphs that thoroughly explore this topic.
+scenario specific texts eg: A developer is tasked with designing an API for a new social media platform. Considering factors like scalability and efficient data fetching, GraphQL would be a suitable choice.
 Include technical definitions, key concepts, practical applications, and any relevant examples.
 make sure the context will be used to genearte mcqs in future so make sure to include context that would yield mcqs that are harder to answer and make the users think thrice before answering a question.
-7. Don't include any special chars like ,/ etc that wouldn't be parsed by python's basic json parser as it would cause an erorr in the system
+Don't include any special chars like ,/ etc that wouldn't be parsed by python's basic json parser as it would cause an erorr in the system
 Make sure to include depth rather than breadth, focusing on the most important aspects of {topics[0]}.
-make sure the context will be used to genearte mcqs in future so make sure to include context that would yield mcqs that are harder to answer and make the users think thrice before answering a question.
-Don't include any special chars like ,/ etc that wouldn't be parsed by python's basic json parser as it would cause an erorr in the system.
 **Instructions for '{difficulty}' difficulty:**
 {difficulty_instruction}
 
@@ -140,6 +139,7 @@ Since there are multiple topics, allocate approximately equal space to each topi
 5. Make sure if the difficulty is hard the context includes super hard insights about the topic i mean hard then comes harder and then comes super hard. 
 6. If the topic includes programming language include some programing snippets but make sure json parser is able to parse it on my end iam using python's basic json parser which may not be able to handle all special characters.
 7. If a topic isn't provided instead a subject or skill is provided like "Javascript" select the topics on your own and generate a relevant context which follows 1 to 6 points as well.  
+scenario specific texts eg: A developer is tasked with designing an API for a new social media platform. Considering factors like scalability and efficient data fetching, GraphQL would be a suitable choice.
 
 The context should still flow as a unified text but ensure balanced coverage of all topics.
 Make sure to include some special application based insights to the context as the context will be used for mcq generation later on so there will be a mix of knowledge and application based mcqs from the context.