C - Context: Provide background information to help LLMs precisely understand the specific scenario, ensuring relevant feedback.
O - Objective: Clearly define the task to enable LLMs to focus responses on achieving that specific goal.
S - Style: Specify the desired writing style, such as a notable figure or industry expert (e.g., business analyst or CEO), guiding LLMs to respond with suitable tone and vocabulary.
T - Tone: Set the emotional tenor to align LLM responses with the intended emotional context, including formal, humorous, empathetic, and more.
A - Audience: Identify the target audience to tailor LLM responses, ensuring content suitability and understandability for experts, beginners, children, etc.
R - Response: Specify the output format (e.g., lists, JSON data, professional reports) to match specific needs, facilitating downstream tasks. JSON is ideal for applications requiring programmatic processing.
System Instructions ( hight level instruction)
Define the operation boundary: Clarify what is not allowed
Define Task
Guide LLM how to format its result.
System instructions act as project briefs for language models, ensuring focused and consistent responses. An usage example is the project instructions for Claude.ai
Setting Params
Elements of a Prompt
Instruction - a specific task or instruction you want the model to perform
Context - external information or additional context that can steer the model to better responses
Input Data - the input or question that we are interested to find a response for
Output Indicator - the type or format of the output.
↓
??Costar framework?
System instructions act as project briefs for language models, ensuring focused and consistent responses. An usage example is the project instructions for Claude.ai