mem-llm 1.0.3__tar.gz → 1.0.4__tar.gz

mem_llm-1.0.4/PKG-INFO

@@ -0,0 +1,304 @@
+Metadata-Version: 2.4
+Name: mem-llm
+Version: 1.0.4
+Summary: Memory-enabled AI assistant with local LLM support
+Home-page: https://github.com/emredeveloper/Mem-LLM
+Author: C. Emre Karataş
+Author-email: karatasqemre@gmail.com
+Project-URL: Bug Reports, https://github.com/emredeveloper/Mem-LLM/issues
+Project-URL: Source, https://github.com/emredeveloper/Mem-LLM
+Keywords: llm ai memory agent chatbot ollama local
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: requests>=2.31.0
+Requires-Dist: pyyaml>=6.0.1
+Provides-Extra: dev
+Requires-Dist: pytest>=7.4.0; extra == "dev"
+Requires-Dist: black>=23.7.0; extra == "dev"
+Requires-Dist: flake8>=6.1.0; extra == "dev"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: keywords
+Dynamic: project-url
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
+
+# 🧠 mem-llm
+
+**Memory-enabled AI assistant that remembers conversations using local LLMs**
+
+[![Python](https://img.shields.io/badge/Python-3.8%2B-blue.svg)](https://www.python.org/downloads/)
+[![PyPI](https://img.shields.io/pypi/v/mem-llm.svg)](https://pypi.org/project/mem-llm/)
+[![License](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
+
+---
+
+## 🎯 What is it?
+
+A lightweight Python library that adds **persistent memory** to local LLM chatbots. Each user gets their own conversation history that the AI remembers across sessions.
+
+**Perfect for:**
+- 💬 Customer service chatbots
+- 🤖 Personal AI assistants  
+- 📝 Context-aware applications
+- 🏢 Business automation
+
+---
+
+## ⚡ Quick Start
+
+### 1. Install
+
+```bash
+pip install mem-llm
+```
+
+### 2. Setup Ollama (one-time)
+
+```bash
+# Install: https://ollama.ai/download
+ollama serve
+
+# Download model (only 2.5GB)
+ollama pull granite4:tiny-h
+```
+
+### 3. Use
+
+```python
+from mem_llm import MemAgent
+
+# Create agent (one line!)
+agent = MemAgent()
+
+# Set user
+agent.set_user("john")
+
+# Chat - it remembers!
+agent.chat("My name is John")
+agent.chat("What's my name?")  # → "Your name is John"
+```
+
+---
+
+## 💡 Features
+
+| Feature | Description |
+|---------|-------------|
+| 🧠 **Memory** | Remembers each user's conversation history |
+| 👥 **Multi-user** | Separate memory for each user |
+| 🔒 **Privacy** | 100% local, no cloud/API needed |
+| ⚡ **Fast** | Lightweight SQLite/JSON storage |
+| 🎯 **Simple** | 3 lines of code to get started |
+
+---
+
+## 📖 Usage Examples
+
+### Basic Chat
+
+```python
+from mem_llm import MemAgent
+
+agent = MemAgent()
+agent.set_user("alice")
+
+# First conversation
+agent.chat("I love pizza")
+
+# Later...
+agent.chat("What's my favorite food?")
+# → "Your favorite food is pizza"
+```
+
+### Customer Service Bot
+
+```python
+agent = MemAgent()
+
+# Customer 1
+agent.set_user("customer_001")
+agent.chat("My order #12345 is delayed")
+
+# Customer 2 (different memory!)
+agent.set_user("customer_002")
+agent.chat("I want to return item #67890")
+```
+
+### Check User Profile
+
+```python
+# Get automatically extracted user info
+profile = agent.get_user_profile()
+# {'name': 'Alice', 'favorite_food': 'pizza', 'location': 'NYC'}
+```
+
+---
+
+## 🔧 Configuration
+
+### JSON Memory (default - simple)
+
+```python
+agent = MemAgent(
+    model="granite4:tiny-h",
+    use_sql=False,  # Use JSON files
+    memory_dir="memories"
+)
+```
+
+### SQL Memory (advanced - faster)
+
+```python
+agent = MemAgent(
+    model="granite4:tiny-h",
+    use_sql=True,  # Use SQLite
+    memory_dir="memories.db"
+)
+```
+
+### Custom Settings
+
+```python
+agent = MemAgent(
+    model="llama2",  # Any Ollama model
+    ollama_url="http://localhost:11434"
+)
+```
+
+---
+
+## 📚 API Reference
+
+### MemAgent
+
+```python
+# Initialize
+agent = MemAgent(model="granite4:tiny-h", use_sql=False)
+
+# Set active user
+agent.set_user(user_id: str, name: Optional[str] = None)
+
+# Chat
+response = agent.chat(message: str, metadata: Optional[Dict] = None) -> str
+
+# Get profile
+profile = agent.get_user_profile(user_id: Optional[str] = None) -> Dict
+
+# System check
+status = agent.check_setup() -> Dict
+```
+
+---
+
+## 🎨 Advanced: PDF/DOCX Config
+
+Generate config from business documents:
+
+```python
+from mem_llm import create_config_from_document
+
+# Create config.yaml from PDF
+create_config_from_document(
+    doc_path="company_info.pdf",
+    output_path="config.yaml",
+    company_name="Acme Corp"
+)
+
+# Use config
+agent = MemAgent(config_file="config.yaml")
+```
+
+---
+
+## 🔥 Models
+
+Works with any [Ollama](https://ollama.ai/) model:
+
+| Model | Size | Speed | Quality |
+|-------|------|-------|---------|
+| `granite4:tiny-h` | 2.5GB | ⚡⚡⚡ | ⭐⭐ |
+| `llama2` | 4GB | ⚡⚡ | ⭐⭐⭐ |
+| `mistral` | 4GB | ⚡⚡ | ⭐⭐⭐⭐ |
+| `llama3` | 5GB | ⚡ | ⭐⭐⭐⭐⭐ |
+
+```bash
+ollama pull <model-name>
+```
+
+---
+
+## 📦 Requirements
+
+- Python 3.8+
+- Ollama (for LLM)
+- 4GB RAM minimum
+- 5GB disk space
+
+**Dependencies** (auto-installed):
+- `requests >= 2.31.0`
+- `pyyaml >= 6.0.1`
+
+---
+
+## 🐛 Troubleshooting
+
+### Ollama not running?
+
+```bash
+ollama serve
+```
+
+### Model not found?
+
+```bash
+ollama pull granite4:tiny-h
+```
+
+### Import error?
+
+```bash
+pip install mem-llm --upgrade
+```
+
+---
+
+## 📄 License
+
+MIT License - feel free to use in personal and commercial projects!
+
+---
+
+## 🔗 Links
+
+- **PyPI:** https://pypi.org/project/mem-llm/
+- **GitHub:** https://github.com/emredeveloper/Mem-LLM
+- **Ollama:** https://ollama.ai/
+
+---
+
+## 🌟 Star us on GitHub!
+
+If you find this useful, give us a ⭐ on [GitHub](https://github.com/emredeveloper/Mem-LLM)!
+
+---
+
+<div align="center">
+Made with ❤️ by <a href="https://github.com/emredeveloper">C. Emre Karataş</a>
+</div>

mem_llm-1.0.4/README.md

@@ -0,0 +1,263 @@
+# 🧠 mem-llm
+
+**Memory-enabled AI assistant that remembers conversations using local LLMs**
+
+[![Python](https://img.shields.io/badge/Python-3.8%2B-blue.svg)](https://www.python.org/downloads/)
+[![PyPI](https://img.shields.io/pypi/v/mem-llm.svg)](https://pypi.org/project/mem-llm/)
+[![License](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
+
+---
+
+## 🎯 What is it?
+
+A lightweight Python library that adds **persistent memory** to local LLM chatbots. Each user gets their own conversation history that the AI remembers across sessions.
+
+**Perfect for:**
+- 💬 Customer service chatbots
+- 🤖 Personal AI assistants  
+- 📝 Context-aware applications
+- 🏢 Business automation
+
+---
+
+## ⚡ Quick Start
+
+### 1. Install
+
+```bash
+pip install mem-llm
+```
+
+### 2. Setup Ollama (one-time)
+
+```bash
+# Install: https://ollama.ai/download
+ollama serve
+
+# Download model (only 2.5GB)
+ollama pull granite4:tiny-h
+```
+
+### 3. Use
+
+```python
+from mem_llm import MemAgent
+
+# Create agent (one line!)
+agent = MemAgent()
+
+# Set user
+agent.set_user("john")
+
+# Chat - it remembers!
+agent.chat("My name is John")
+agent.chat("What's my name?")  # → "Your name is John"
+```
+
+---
+
+## 💡 Features
+
+| Feature | Description |
+|---------|-------------|
+| 🧠 **Memory** | Remembers each user's conversation history |
+| 👥 **Multi-user** | Separate memory for each user |
+| 🔒 **Privacy** | 100% local, no cloud/API needed |
+| ⚡ **Fast** | Lightweight SQLite/JSON storage |
+| 🎯 **Simple** | 3 lines of code to get started |
+
+---
+
+## 📖 Usage Examples
+
+### Basic Chat
+
+```python
+from mem_llm import MemAgent
+
+agent = MemAgent()
+agent.set_user("alice")
+
+# First conversation
+agent.chat("I love pizza")
+
+# Later...
+agent.chat("What's my favorite food?")
+# → "Your favorite food is pizza"
+```
+
+### Customer Service Bot
+
+```python
+agent = MemAgent()
+
+# Customer 1
+agent.set_user("customer_001")
+agent.chat("My order #12345 is delayed")
+
+# Customer 2 (different memory!)
+agent.set_user("customer_002")
+agent.chat("I want to return item #67890")
+```
+
+### Check User Profile
+
+```python
+# Get automatically extracted user info
+profile = agent.get_user_profile()
+# {'name': 'Alice', 'favorite_food': 'pizza', 'location': 'NYC'}
+```
+
+---
+
+## 🔧 Configuration
+
+### JSON Memory (default - simple)
+
+```python
+agent = MemAgent(
+    model="granite4:tiny-h",
+    use_sql=False,  # Use JSON files
+    memory_dir="memories"
+)
+```
+
+### SQL Memory (advanced - faster)
+
+```python
+agent = MemAgent(
+    model="granite4:tiny-h",
+    use_sql=True,  # Use SQLite
+    memory_dir="memories.db"
+)
+```
+
+### Custom Settings
+
+```python
+agent = MemAgent(
+    model="llama2",  # Any Ollama model
+    ollama_url="http://localhost:11434"
+)
+```
+
+---
+
+## 📚 API Reference
+
+### MemAgent
+
+```python
+# Initialize
+agent = MemAgent(model="granite4:tiny-h", use_sql=False)
+
+# Set active user
+agent.set_user(user_id: str, name: Optional[str] = None)
+
+# Chat
+response = agent.chat(message: str, metadata: Optional[Dict] = None) -> str
+
+# Get profile
+profile = agent.get_user_profile(user_id: Optional[str] = None) -> Dict
+
+# System check
+status = agent.check_setup() -> Dict
+```
+
+---
+
+## 🎨 Advanced: PDF/DOCX Config
+
+Generate config from business documents:
+
+```python
+from mem_llm import create_config_from_document
+
+# Create config.yaml from PDF
+create_config_from_document(
+    doc_path="company_info.pdf",
+    output_path="config.yaml",
+    company_name="Acme Corp"
+)
+
+# Use config
+agent = MemAgent(config_file="config.yaml")
+```
+
+---
+
+## 🔥 Models
+
+Works with any [Ollama](https://ollama.ai/) model:
+
+| Model | Size | Speed | Quality |
+|-------|------|-------|---------|
+| `granite4:tiny-h` | 2.5GB | ⚡⚡⚡ | ⭐⭐ |
+| `llama2` | 4GB | ⚡⚡ | ⭐⭐⭐ |
+| `mistral` | 4GB | ⚡⚡ | ⭐⭐⭐⭐ |
+| `llama3` | 5GB | ⚡ | ⭐⭐⭐⭐⭐ |
+
+```bash
+ollama pull <model-name>
+```
+
+---
+
+## 📦 Requirements
+
+- Python 3.8+
+- Ollama (for LLM)
+- 4GB RAM minimum
+- 5GB disk space
+
+**Dependencies** (auto-installed):
+- `requests >= 2.31.0`
+- `pyyaml >= 6.0.1`
+
+---
+
+## 🐛 Troubleshooting
+
+### Ollama not running?
+
+```bash
+ollama serve
+```
+
+### Model not found?
+
+```bash
+ollama pull granite4:tiny-h
+```
+
+### Import error?
+
+```bash
+pip install mem-llm --upgrade
+```
+
+---
+
+## 📄 License
+
+MIT License - feel free to use in personal and commercial projects!
+
+---
+
+## 🔗 Links
+
+- **PyPI:** https://pypi.org/project/mem-llm/
+- **GitHub:** https://github.com/emredeveloper/Mem-LLM
+- **Ollama:** https://ollama.ai/
+
+---
+
+## 🌟 Star us on GitHub!
+
+If you find this useful, give us a ⭐ on [GitHub](https://github.com/emredeveloper/Mem-LLM)!
+
+---
+
+<div align="center">
+Made with ❤️ by <a href="https://github.com/emredeveloper">C. Emre Karataş</a>
+</div>

{mem_llm-1.0.3 → mem_llm-1.0.4}/mem_llm/__init__.py

@@ -24,7 +24,7 @@ try:
 except ImportError:
     __all_pro__ = []
 
-__version__ = "1.0.3"
+__version__ = "1.0.4"
 __author__ = "C. Emre Karataş"
 
 __all__ = [

{mem_llm-1.0.3 → mem_llm-1.0.4}/mem_llm/mem_agent.py

@@ -220,22 +220,30 @@ class MemAgent:
             except Exception as e:
                 self.logger.error(f"Prompt template loading error: {e}")
                 # Simple, short and effective default prompt
-                self.current_system_prompt = """You are a concise AI assistant. Be EXTREMELY brief.
+                self.current_system_prompt = """You are a helpful AI assistant with access to a knowledge base.
 
-RULES (MANDATORY):
-1. MAX 1-2 SHORT sentences per response
-2. When user shares info: Just say "Got it!" or "Noted!"
-3. Answer questions: ONE sentence, direct
-4. NO lists, NO explanations, NO examples
-5. Use conversation history when relevant
+CRITICAL RULES (FOLLOW EXACTLY):
+1. If KNOWLEDGE BASE information is provided below, USE IT FIRST - it's the correct answer!
+2. Knowledge base answers are marked with "📚 RELEVANT KNOWLEDGE BASE"
+3. Keep responses SHORT (1-3 sentences maximum)
+4. When user shares personal info: Just acknowledge briefly ("Got it!" or "Noted!")
+5. Answer from knowledge base EXACTLY as written, don't make up information
+6. If knowledge base has no info, use conversation history or say "I don't have that information"
+
+RESPONSE PRIORITY:
+1st Priority: Knowledge Base (if available) ← USE THIS!
+2nd Priority: Conversation History
+3rd Priority: General knowledge (be brief)
 
 EXAMPLES:
+User: "What's the shipping cost?"
+Knowledge Base: "Shipping is free over $150"
+You: "Shipping is free for orders over $150!"
+
 User: "My name is Alice" → You: "Nice to meet you, Alice!"
-User: "My favorite food is pizza" → You: "Got it!"
 User: "What's my name?" → You: "Your name is Alice."
-User: "Tell me about Python" → You: "Python is a versatile programming language for web, data science, and AI."
 
-BE BRIEF OR USER WILL LEAVE!"""
+REMEMBER: Knowledge base = truth. Always use it when provided!"""
 
     def check_setup(self) -> Dict[str, Any]:
         """Check system setup"""
@@ -322,18 +330,24 @@ BE BRIEF OR USER WILL LEAVE!"""
 
         # Knowledge base search (if using SQL)
         kb_context = ""
-        if ADVANCED_AVAILABLE and isinstance(self.memory, SQLMemoryManager) and hasattr(self, 'config') and self.config:
-            if self.config.get("response.use_knowledge_base", True):
+        if ADVANCED_AVAILABLE and isinstance(self.memory, SQLMemoryManager):
+            # Check config only if it exists, otherwise always use KB
+            use_kb = True
+            kb_limit = 5
+            
+            if hasattr(self, 'config') and self.config:
+                use_kb = self.config.get("response.use_knowledge_base", True)
+                kb_limit = self.config.get("knowledge_base.search_limit", 5)
+            
+            if use_kb:
                 try:
-                    kb_results = self.memory.search_knowledge(
-                        query=message,
-                        limit=self.config.get("knowledge_base.search_limit", 5)
-                    )
+                    kb_results = self.memory.search_knowledge(query=message, limit=kb_limit)
 
                     if kb_results:
-                        kb_context = "\n\nRelevant Information:\n"
+                        kb_context = "\n\n📚 RELEVANT KNOWLEDGE BASE:\n"
                         for i, result in enumerate(kb_results, 1):
-                            kb_context += f"{i}. S: {result['question']}\n   C: {result['answer']}\n"
+                            kb_context += f"{i}. Q: {result['question']}\n   A: {result['answer']}\n"
+                        kb_context += "\n⚠️ USE THIS INFORMATION TO ANSWER! Be brief but accurate.\n"
                 except Exception as e:
                     self.logger.error(f"Knowledge base search error: {e}")
 
@@ -355,15 +369,13 @@ BE BRIEF OR USER WILL LEAVE!"""
         except Exception as e:
             self.logger.error(f"Memory history loading error: {e}")
 
-        # Add knowledge base context
+        # Add current message WITH knowledge base context (if available)
+        final_message = message
         if kb_context:
-            messages.append({
-                "role": "system",
-                "content": f"You can use this information when answering the user's question:{kb_context}"
-            })
-
-        # Add current message
-        messages.append({"role": "user", "content": message})
+            # Inject KB directly into user message for maximum visibility
+            final_message = f"{kb_context}\n\nUser Question: {message}"
+        
+        messages.append({"role": "user", "content": final_message})
 
         # Get response from LLM
         try: