Name AI assistant Hesychia in system prompts (#20)

Closes #20

Introduce a shared assistant identity module and prepend it to
user-facing generation prompts so the model self-identifies as Hesychia.

README.md

@@ -95,6 +95,10 @@ with `COMPOSE_DATABASE_URL` if needed.
 | `GUNICORN_WORKERS` / `GUNICORN_BIND` | 2 / `0.0.0.0:8000` | optional | Entrypoint |
 | `SKIP_RAG_INIT` | unset | CI/migrate often `1` | Skip Chroma/Ollama boot work |
 
+Assistant identity (`Hesychia`) lives in code:
+`llm_be/chat_backend/services/assistant_identity.py` — prepended to user-facing
+generation prompts (chat, RAG, data analysis). Not env-configurable.
+
 Templates: `.env.example` (local), `.env.prod.example` (control-node secret).
 
 Control-node secret path (server-infra on ai-server-4080):

llm_be/chat_backend/services/assistant_identity.py

@@ -0,0 +1,13 @@
+"""Assistant identity for user-facing LLM prompts.
+
+Hesychia is the product assistant name (hesychia.ai). Keep this concise —
+it is prepended to generation prompts, not classifiers/moderators/title gens.
+"""
+
+ASSISTANT_NAME = "Hesychia"
+
+ASSISTANT_SYSTEM_PROMPT = (
+    "You are Hesychia, a helpful AI assistant. "
+    "Your name evokes quiet, rest, silence, and stillness — "
+    "respond with calm clarity; keep answers focused and uncluttered."
+)

llm_be/chat_backend/services/data_analysis_service.py

@@ -12,6 +12,7 @@ import docx
 import pypdf
 from django.conf import settings
 from chat_backend.ollama_config import ollama_llm_kwargs
+from chat_backend.services.assistant_identity import ASSISTANT_SYSTEM_PROMPT
 
 
 class AsyncDataAnalysisService:
@@ -30,7 +31,8 @@ class AsyncDataAnalysisService:
 
     def _setup_chain(self):
         """Set up the LLM chain with a prompt tailored for data analysis."""
-        template = """You are an expert data analyst. Your role is to directly answer a user's question about a dataset or document they have provided.
+        template = f"""{ASSISTANT_SYSTEM_PROMPT}
+For this request, act as an expert data analyst. Your role is to directly answer a user's question about a dataset or document they have provided.
 You will be given a summary and a sample of the dataset, or the content of the document.
 Based on this information, provide a clear and concise answer to the user's question.
 Do not provide Python code or any other code. The user is not a developer and wants a direct answer.
@@ -38,10 +40,10 @@ Even if you don't think the data provides enough evidence for the query, still p
 
 ---
 Data/Document Content:
-{data_summary}
+{{data_summary}}
 ---
 
-User's Question: {query}
+User's Question: {{query}}
 Answer:"""
 
         self.prompt = ChatPromptTemplate.from_template(template)

llm_be/chat_backend/services/llm_service.py

@@ -9,6 +9,7 @@ from django.conf import settings
 
 from chat_backend.models import Conversation, Prompt
 from chat_backend.ollama_config import ollama_llm_kwargs
+from chat_backend.services.assistant_identity import ASSISTANT_SYSTEM_PROMPT
 
 
 class LLMService(ABC):
@@ -50,11 +51,13 @@ class SyncLLMService(LLMService):
 
     def _setup_chain(self):
         """Setup the conversation chain."""
-        template = """Continue the conversation based on the following history:
+        template = f"""{ASSISTANT_SYSTEM_PROMPT}
+
+        Continue the conversation based on the following history:
         
-        {history}
+        {{history}}
         
-        Latest message: {query}
+        Latest message: {{query}}
         
         Response:"""
         self.prompt = ChatPromptTemplate.from_template(template)
@@ -88,13 +91,15 @@ class AsyncLLMService(LLMService):
 
     def _setup_chain(self):
         """Setup the conversation chain."""
-        template = """Continue this conversation while maintaining context by providing a single helpful response. 
+        template = f"""{ASSISTANT_SYSTEM_PROMPT}
-        Current context: {context}
+
+        Continue this conversation while maintaining context by providing a single helpful response. 
+        Current context: {{context}}
         
         Last 3 messages:
-        {recent_history}
+        {{recent_history}}
         
-        Latest message: {query}
+        Latest message: {{query}}
         
         Instructions:
         - Carefully maintain all established context

llm_be/chat_backend/services/rag_services.py

@@ -24,6 +24,7 @@ from django.core.files.uploadedfile import UploadedFile
 from chat_backend.models import Conversation, Prompt, DocumentWorkspace, Document
 from pathlib import Path
 from chat_backend.services.base_service import BaseService
+from chat_backend.services.assistant_identity import ASSISTANT_SYSTEM_PROMPT
 from chat_backend.ollama_config import ollama_embeddings_kwargs
 
 
@@ -229,13 +230,15 @@ class SyncRAGService(RAGService):
 
     def _setup_chain(self):
         """Setup the RAG chain."""
-        template = """Answer the question based only on the following context:
+        template = f"""{ASSISTANT_SYSTEM_PROMPT}
-        {context}
+
+        Answer the question based only on the following context:
+        {{context}}
         
         Conversation history:
-        {history}
+        {{history}}
         
-        Question: {question}
+        Question: {{question}}
         """
         self.prompt = ChatPromptTemplate.from_template(template)
 
@@ -305,13 +308,15 @@ class AsyncRAGService(RAGService):
 
     def _setup_chain(self):
         """Setup the RAG chain."""
-        template = """Answer the question based only on the following context:
+        template = f"""{ASSISTANT_SYSTEM_PROMPT}
-        {context}
+
+        Answer the question based only on the following context:
+        {{context}}
         
         Conversation history:
-        {history}
+        {{history}}
         
-        Question: {question}
+        Question: {{question}}
         """
         self.prompt = ChatPromptTemplate.from_template(template)
 

llm_be/chat_backend/tests/test_assistant_identity.py

@@ -0,0 +1,28 @@
+from django.test import SimpleTestCase
+
+from chat_backend.services.assistant_identity import (
+    ASSISTANT_NAME,
+    ASSISTANT_SYSTEM_PROMPT,
+)
+from chat_backend.services.data_analysis_service import AsyncDataAnalysisService
+from chat_backend.services.llm_service import AsyncLLMService, SyncLLMService
+
+
+class AssistantIdentityTestCase(SimpleTestCase):
+    def test_assistant_is_named_hesychia(self):
+        self.assertEqual(ASSISTANT_NAME, "Hesychia")
+        self.assertIn("Hesychia", ASSISTANT_SYSTEM_PROMPT)
+        self.assertNotRegex(
+            ASSISTANT_SYSTEM_PROMPT,
+            r"(?i)\b(chatgpt|claude|gemini|copilot)\b",
+        )
+
+    def test_user_facing_prompts_include_hesychia(self):
+        services = [
+            SyncLLMService(),
+            AsyncLLMService(),
+            AsyncDataAnalysisService(),
+        ]
+        for service in services:
+            with self.subTest(service=type(service).__name__):
+                self.assertIn("Hesychia", str(service.prompt))

llm_be/chat_backend/views.py

@@ -44,6 +44,7 @@ import json
 import base64
 import pandas as pd
 import io
+from chat_backend.services.assistant_identity import ASSISTANT_SYSTEM_PROMPT
 
 # For email support
 from django.core.mail import EmailMultiAlternatives
@@ -714,7 +715,7 @@ class AdminAnalytics(APIView):
 
 
 prompt = ChatPromptTemplate.from_messages(
-    [("system", "You are a helpful assistant."), ("user", "{input}")]
+    [("system", ASSISTANT_SYSTEM_PROMPT), ("user", "{input}")]
 )
 
 llm = OllamaLLM(**ollama_llm_kwargs(model=MODEL_NAME))