mem-llm 1.0.7__tar.gz → 1.0.10__tar.gz

mem_llm-1.0.10/CHANGELOG.md

@@ -0,0 +1,198 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [1.0.10] - 2025-10-20
+
+### Added
+- 🧠 **Dynamic Prompt System**: Context-aware system prompts that adapt to active features
+  - Prevents hallucinations by only including instructions for enabled features
+  - Separate prompt sections for KB, tools, business/personal modes
+  - Automatic feature detection (Knowledge Base presence, tools availability)
+  - Logging shows active features: "✅ Knowledge Base | ❌ Tools | 💾 Memory: SQL"
+- 🔄 **Universal Ollama Model Compatibility**: Full support for ALL Ollama models
+  - Thinking-enabled models (Qwen3, DeepSeek, etc.) now work correctly
+  - Auto-detection and handling of thinking mode
+  - `enable_thinking: false` parameter for direct responses
+  - Fallback extraction from thinking process when needed
+  - Empty response retry with simpler prompts
+- 📊 **Comprehensive Test Suite**: Pre-publish validation system
+  - 34 automated tests covering all major features
+  - Tests: imports, CLI, Ollama, JSON/SQL memory, MemAgent, config, multi-user
+  - User scenario testing with output analysis
+  - Hallucination detection and context verification
+
+### Changed
+- ⚡ **LLM Token Limits**: Increased from 150 to 2000 tokens for thinking models
+- 🧹 **Removed Obsolete Module**: Deleted `prompt_templates.py` (replaced by dynamic system)
+- 📝 **Context Window**: Increased from 2048 to 4096 tokens for better context
+- 🎯 **Response Quality**: Better handling of empty responses with automatic retry
+
+### Fixed
+- 🐛 **Thinking Model Issue**: Qwen3 and similar models now respond correctly
+  - Fixed empty responses from thinking-mode models
+  - Proper content extraction from model responses
+  - System prompt instructions to suppress thinking process
+- 🔧 **Stop Sequences**: Removed problematic stop sequences that interfered with models
+- ⚠️ **Empty Response Handling**: Automatic retry with fallback for reliability
+
+### Improved
+- 🎨 **Prompt Quality**: Feature-specific instructions prevent confusion
+- 🚀 **Model Performance**: Works seamlessly with granite4, qwen3, llama3, and all Ollama models
+- 📈 **User Experience**: No more irrelevant feature mentions in responses
+- 🧪 **Testing Coverage**: Complete validation before releases
+
+### Technical Details
+- Created `mem_llm/dynamic_prompt.py` (350+ lines) - modular prompt builder
+- Modified `mem_llm/mem_agent.py`:
+  - Added `has_knowledge_base` and `has_tools` tracking flags
+  - Implemented `_build_dynamic_system_prompt()` method
+  - Removed ~70 lines of old static prompt code
+  - Added empty response retry logic
+- Modified `mem_llm/llm_client.py`:
+  - Added thinking mode detection and suppression
+  - Increased token limits and context window
+  - Improved response extraction logic
+  - Added fallback for thinking-enabled models
+- Updated `mem_llm/__init__.py` - exported `dynamic_prompt_builder`
+- Cleaned `MANIFEST.in` - removed non-existent files
+- Created `comprehensive_test.py` - 34 automated tests
+- Created `user_test.py` - real-world scenario validation
+
+### Breaking Changes
+- None - fully backward compatible
+
+## [1.0.9] - 2025-10-20
+
+### Added
+- 📝 **PyPI-Optimized README**: Complete rewrite with practical examples
+  - 5 comprehensive usage examples with full code and output
+  - Print statements in all examples for better user experience
+  - Step-by-step workflows showing complete processes
+  - Real-world customer service scenario example
+  - Turkish language support demonstration
+  - User profile extraction example
+- 📄 **Document Configuration Examples**: Added example demonstrating PDF/DOCX/TXT config generation
+- 🧪 **Config Update Testing**: Verification that manual YAML edits work correctly
+
+### Changed
+- 🗑️ **Removed docs folder**: Consolidated documentation into main README
+- 🪵 **Logging Behavior**: Changed from file+console to console-only logging
+  - No more `mem_agent.log` files cluttering workspace
+  - Keeps workspace clean with only `.db` and `.yaml` files
+- 📖 **Example Format**: All examples now include:
+  - Print statements for visibility
+  - Expected output blocks
+  - Full conversation flows
+  - Real usage scenarios
+
+### Fixed
+- 🐛 **Log File Pollution**: Removed FileHandler from logging, only StreamHandler now
+- 📝 **README Examples**: Fixed examples that didn't show actual output or complete process
+
+### Improved
+- 🎯 **User Experience**: Much clearer examples for new users
+- 📚 **Documentation Quality**: Professional PyPI-ready documentation
+- 🔍 **Example Clarity**: Each example shows input, process, and output
+
+### Technical Details
+- Modified `mem_llm/mem_agent.py` - removed FileHandler from logging setup
+- Rewrote `README.md` with 5 detailed examples
+- Created `examples/07_document_config.py` for PDF/DOCX/TXT feature
+- Verified config changes work correctly with manual YAML edits
+
+## [1.0.8] - 2025-10-20
+
+### Added
+- 🎯 **CLI Tool**: Full-featured command-line interface
+  - `mem-llm chat` - Interactive chat sessions
+  - `mem-llm check` - System verification
+  - `mem-llm stats` - Statistics and analytics
+  - `mem-llm export` - Data export (JSON/TXT)
+  - `mem-llm clear` - User data deletion
+- 📊 **Feature Comparison Matrix**: Clear comparison between JSON and SQL modes
+- 📦 **Improved Dependencies**: Proper separation of core, dev, and optional requirements
+  - `requirements.txt` - Core dependencies only
+  - `requirements-dev.txt` - Development tools
+  - `requirements-optional.txt` - Optional features (web, API, etc.)
+- 🔧 **Better Error Handling**: Improved startup checks with user-friendly messages
+- 📚 **Enhanced Documentation**: CLI usage examples and feature matrices
+
+### Changed
+- 🌍 **Multi-language Support**: Changed from "Turkish Support" to general multi-language
+- 📖 **Documentation**: All content now in English for broader accessibility
+- 🎨 **CLI Entry Point**: Added `mem-llm` console script in setup.py
+
+### Fixed
+- 🐛 Missing `click` dependency in requirements
+- 🐛 Improved error messages when Ollama is not running
+
+### Improved
+- ⚡ Better user experience with CLI commands
+- 📝 Clearer README with usage examples
+- 🎯 More intuitive API design
+
+## [1.0.4] - 2025-10-13
+
+### Added
+- ✨ Config-free knowledge base support - KB now works without config.yaml
+- ✨ Smart keyword extraction for knowledge base search (Turkish & English stopwords)
+- ✨ Enhanced KB context injection - KB data injected directly into user message
+- ✨ Automatic user profile extraction (name, favorite_food, location)
+- ✨ Turkish language support for profile extraction
+- ✨ SQL-JSON memory compatibility methods
+- 📚 New example: `example_knowledge_base.py`
+- 🧪 Comprehensive test suite
+
+### Fixed
+- 🐛 Knowledge base not being used without config.yaml
+- 🐛 LLM ignoring knowledge base information
+- 🐛 User profiles returning empty dictionaries
+- 🐛 Profile updates not working correctly with SQL memory
+- 🐛 Keyword search failing with Turkish queries
+
+### Improved
+- ⚡ Better KB-first response priority in system prompts
+- ⚡ More accurate answers from knowledge base
+- ⚡ Enhanced search algorithm with stopword filtering
+
+## [1.0.3] - 2025-10-12
+
+### Added
+- 📦 Initial PyPI release
+- 🎯 Core memory features (JSON & SQL)
+- 🤖 Ollama integration
+- 💾 Knowledge base system
+- 🛠️ User tools
+- ⚙️ Configuration management
+
+### Features
+- Memory-enabled AI agent
+- JSON and SQL memory backends
+- Knowledge base integration
+- User profile management
+- Conversation history
+- Configuration from YAML/documents
+
+## [1.0.2] - 2025-10-11
+
+### Internal
+- 🔧 Package structure improvements
+- 📝 Documentation updates
+
+## [1.0.1] - 2025-10-10
+
+### Fixed
+- 🐛 Import errors after package rename
+- 📦 Package directory naming issues
+
+## [1.0.0] - 2025-10-09
+
+### Initial Release
+- 🎉 First stable release
+- 🤖 Memory-enabled AI assistant
+- 💾 JSON memory management
+- 🔌 Ollama integration

{mem_llm-1.0.7 → mem_llm-1.0.10}/MANIFEST.in

@@ -1,11 +1,8 @@
 include README.md
 include requirements.txt
+include requirements-dev.txt
+include requirements-optional.txt
 include CHANGELOG.md
-include QUICKSTART.md
-include QUICKSTART_TR.md
-include INTEGRATION_GUIDE.md
-include STRUCTURE.md
 recursive-include mem_llm *.yaml
 recursive-include mem_llm *.example
-recursive-include docs *.md