A small interactive CLI tool that uses LLM prompt-chains to diagnose student responses to short science concept questions and provide kind, targeted instructional feedback, including a short follow-up (mastery) check. Intended for teachers who want specific feedback, including:
feedback to students,
follow-up clarifying questions, and
generating conceptual questions about a science topic.
The teacher either:
Select a topic and the Agent generates 5 problems for that particular question, or
Enter their own question, if they have one
The program will ask teachers to enter students' responses to give feedback.
The agent provides:
Diagnosis - the initial diagnosis of the student's understanding, whether the student's answer is missing anything.
Feedback - a feedback written in the teacher's tone, very kind and encouraging.
Micro-activity - an extension of the question with some small hands-on projects that the student can try out themselves to deepen their understanding of the topic.
Teacher note - tailored instructional advice to this specific student.
Clarifying questions and answers - gives a follow-up question that the teacher can give to the student to clarify the student's current understanding.
Mastery check - provides a series of information:
The clarifying question,
They expected key points in the student's response, and
Common wrong signals that the teacher needs to be aware of.
Then the tool prompts the teacher to enter student response to the follow-up question.
The agent gives the student a final feedback, considering both questions and answers from round 1 and round 2, to reach a final conclusion about the student's current knowledge and lack of understanding.
It also provides a micro-activity, as well as a teacher's note, focused on these overall insights.