- Add is_summarized boolean field to ToolCallResult schema
- Implement logic to mark tool calls as summarized after processing
- Skip summarization for tools with all recent calls already summarized
- Update summary operation to process only tools with unsummarized calls
- Enhance logging to show summarization and skip statistics
- Modify response to include count of summarized vs skipped tools
- Add documentation for smart skip logic and is_summarized feature
- Include test scenarios for skip logic and incremental summarization
- Enhanced tool call result parsing with improved scoring logic (0.0 or 1.0)
- Updated tool memory schema to reflect binary success/failure scoring
- Added deterministic behavior support via seed configuration in mock tools
- Improved evaluation prompts to focus on result quality over success flags
- Extended README with tool memory documentation and usage examples- Added utility functions for generating mock tool call results
- Removed deprecated test file for UseMockSearchOp- Updated default configurations to include use_mock_search operation- Bumped version to0.1.10 and updated flowllm dependency requirement
- Moved deprecation warnings to main init file
- Simplified tool memory summary formatting by removing redundant statistics
- Fixed tool call result processing to handle multiple tool names concurrently
- Added LLMMockSearchOp for simulating search operations with configurable complexity- Created SearchToolA, SearchToolB, and SearchToolC with distinct performance profiles- Implemented UseMockSearchOp for intelligent tool selection based on query analysis
- Added test scripts and query datasets for evaluating tool memory effectiveness
- Integrated tool memory service for storing and retrieving tool performance data
- Created documentation for tool memory benchmark testing methodology
- Refactored agent module structure and imports
- Increased summary tool memory recent call count from20 to 30
- Added SummaryToolMemoryOp for analyzing tool usage patterns- Implemented RetrieveToolMemoryOp for fetching tool memories
- Enhanced ParseToolCallResultOp with evaluation and scoring logic
- Updated ToolMemory schema with summary, evaluation, and score fields
- Added new prompt templates for tool evaluation and summarization
- Configured tool memory operations in default.yaml
- Fixed success flag logic in ToolCallResult processing
- Improved statistical analysis for tool call history
- Added concurrent processing for tool call evaluations
- Integrated tool memory updates with vector store operations
- Add new ToolCallResult and ToolMemory schemas for tracking tool executions
- Implement tool memory retrieval and summarization flows in default config
- Register new parse_tool_call_result_op for processing tool call results
- Extend memory conversion logic to support tool memory type
- Add test cases for tool memory serialization and deserialization
- Include token counting utilities for text processing tasks
- Remove deprecated `add_dict_filter` method from vector store operations
- Update `async_search` method to use `filter_dict` parameter for advanced filtering
- Modify examples in documentation to use the new `filter_dict` format for advanced filtering
- Update task memory and messages for consistency with new filtering approach
- Update personal memory demo to use new memory format
- Remove unused task memory and message files
- Adjust vector store configuration in default.yaml
- Modify load_today_memory_op to use new search method
- Update load_today_memory_op.py to use async version of _retrieve_today_memories
- Update memory_deduplication_op.py to use async version of _deduplicate_task_memories and _get_existing_task_memory_embeddings
- Update retrieve_memory_op.py to use async_search instead of search
- Modify LLM-related ops to use async_execute instead of execute
- Replace chat method with achat for asynchronous LLM calls
- Update method signatures and return types to support async operations
- This change affects multiple files across the project
- Update app.py to use async service initialization
- Refactor multiple ops to use async_execute instead of execute
- Add support for stream and use_async flags in config
- Update LLM usage to use achat instead of chat
- Add new LLM models and update existing ones in config
- Improve error handling and logging in several ops
- Update dependencies and Python version requirements
- Remove unused import in reme_ai/summary/task/__init__.py
- Remove unused constant in reme_ai/constants/common_constants.py
- Update README.md with new content and structure
- Update README_ZH.md with minor corrections
- Modify task memory documentation for clarity
- Move retrieve_personal_memory and summary_personal_memory flows to new positions in default.yaml
- Update info_filter_op to handle trajectories instead of messages
- Adjust use_personal_memory_demo to use trajectories in API requests
- Added 'import json' at the beginning of trajectory_preprocess_op.py
- This import is necessary for handling JSON data in the trajectory preprocessing task
- Add SimpleReactOp to react module
- Update reme_ai/__init__.py to include react module
- Modify contra_repeat_op.py to use memory_id instead of id
- Adjust datetime_handler.py to handle string datetime conversion
- Update default.yaml to include react flow content
- Modify get_observation_op.py and get_observation_with_time_op.py to use workspace_id from context
- Update test/http_client_test.py to test new react functionality
- Adjust messages.jsonl to reflect new analysis approach for Xiaomi Corporation
- Rename and restructure personal memory consolidation flow
- Update memory handling in context and operations
- Refactor memory schema to use time_created and time_modified fields
- Improve error handling and logging in memory operations
- Update test cases for new memory consolidation flow
- Rename and restructure memory loading operations
- Enhance time extraction and formatting capabilities- Implement more robust memory filtering and ranking logic
- Refactor observation extraction methods for better reusability
- Improve logging and error handling in memory operations
- Rename `worker` to `op` for operation classes
- Remove `memoryscope` package
- Create `constants` package with common constants
- Update module imports and class names accordingly
- Add agent module to reme_ai package
- Update retrieve module to include agent-related functionality
- Modify summary module to use new agent features
- Update configuration to support agent operations
- Update import statements to use relative imports where appropriate
- Rename classes to match file names for better consistency
- Add new imports for recently added modules
- Rename ExperienceDeduplicationOp to TaskMemoryDeduplicationOp
- Rename ExperienceValidationOp to TaskMemoryValidationOp
- Update comparative extraction, failure extraction, and success extraction ops to use task memories
- Refactor PDFPreprocessOp and ReactV1Op for better code structure
- Remove unused simple_config.yaml
- Update default.yaml with new task memory related flows
- Add summary module with simple summary and personal summary tasks
- Implement new vector store operations for updating and managing memory
- Refactor retrieve module to improve query building and memory rewriting
- Update schema and utils modules for better data handling and PDF processing
- Rename experiencemaker package to reme_ai
- Move personal modules to new directory structure
- Remove unused classes and imports
- Update module initialization files
- Add BuildQueryOp to construct query for memory retrieval-Implement MergeMemoryOp to combine retrieved memories
- Create RecallVectorStoreOp to fetch memories from vector store
- Develop memory representation and conversion methods
- Establish initial project structure and dependencies
