update docs

Author: neagualexa

docs/dev.md

@@ -3,13 +3,24 @@ This chatbot aims to concisely respond to all relevant tasks the student request
 
 Some technical details:
 <pre style="white-space: pre-wrap;">
-<code>LLM model: Gemini-2.5-flash (Google)
-response time (on average): 10 seconds
+<code>LLM model: Gemini-2.5-flash-lite (Google)
+response time (on average): ~10 seconds
 
 Helping approach: provides short, direct answers without extra detail
 </code>
 </pre>
 
+## Behavioural Prompt
+
+**Role:** A tutor that gives short, direct answers.
+
+**Key behaviours from the role prompt**
+
+- **Direct and minimal:** answers the question and stops — no extra details, explanations, or examples unless the student asks.
+- **Aware of struggle:** if the student seems stuck or frustrated, references their progress so far and how long they have spent on the question relative to the teacher's guidance time.
+- **Stays on topic:** redirects unrelated questions back to the current material with a short refusal.
+- **No filler:** does not end messages with concluding statements or summaries.
+
 ## Testing the Chat Function
 
 To test your function, you can run the unit tests, call the code directly through a python script, or build the respective chat function docker container locally and call it through an API request. Below you can find details on those processes.

docs/user.md

@@ -1,12 +1,19 @@
 # Concise Chatbot
 
-This chatbot aims to concisely respond to all relevant tasks the student requests based on the current Question they are working on. The Chatbot is aware of the Question details, answer, worked solution and guidance from the lecturer.
+Short, direct answers and nothing else. No preamble, no examples, no details — unless you ask for them. Best when you just need one missing fact to get unstuck.
+
+**Example chat**
+
+> **You:** How can I start solving this question?
+>
+> **Concise:** Consider which governing equation applies to this problem.
+
+**Tip:** if a hint is too short, ask it to explain *why* — it'll expand.
+
 
 Some technical details:
 <pre style="white-space: pre-wrap;">
-<code>LLM model: Gemini-2.0-flash (Google)
-response time (on average): 10 seconds
-
-Helping approach: provides short, direct answers without extra detail
+<code>LLM model: Gemini-2.5-flash-lite (Google)
+response time (on average): ~10 seconds
 </code>
 </pre>