v1.18.6 add intelligent recording guidance to record_context tool

Enhanced

• Intelligent Recording Guidance: Enhanced record_context tool with smart usage guidelines
  • Added "When to Record" section: record after user confirmation, successful solutions, completed features
  • Added "When NOT to Record" section: avoid recording during exploration, failed attempts, temporary states
  • Emphasized automatic quality filtering (quality_score < 0.6)
  • Helps AI make better decisions about when to record context
  • Reduces noise and improves memory quality

Changed

• MCP Config Example: Updated autoApprove list to include record_context
  • Enables automatic memory recording for MCP client users (who don't have daemon)
  • Balances convenience with quality through prompt-based guidance
  • Users can manually remove from autoApprove if they prefer manual approval

Technical Details

• Implemented combination approach: Prompt guidance + Quality filtering
• No code logic changes, only tool description enhancement
• Maintains backward compatibility

v1.18.5 clarify project_analysis_engineer prompt file naming instructions

Fixed

• project_analysis_engineer Prompt Clarity: Fixed ambiguous file naming instructions
  • Changed from "generate DEVMIND.md style document" to "generate or update DEVMIND.md file"
  • Added explicit instructions: file name MUST be DEVMIND.md (no version suffixes)
  • Specified incremental update behavior: update existing file instead of creating new ones
  • Added instruction to preserve valuable existing content
  • Applied same fixes to both DEVMIND.md and CLAUDE.md generation
  • Prevents AI from creating duplicate files like DEVMIND-v2.md

Technical Details

• Enhanced prompt instructions with clear file naming rules
• Added bilingual support (Chinese/English) for all instructions
• Improved documentation generation consistency

v1.18.4 Enhanced memory graph tooltip interaction and layout optimization

Enhanced

• Memory Graph Tooltip Interaction: Significantly improved tooltip usability
  • Added hover state tracking to prevent accidental tooltip triggers
  • Tooltip only appears when mouse is truly over a node
• Mouse can smoothly move from node to tooltip for scrolling long content
  • Intelligent positioning to prevent tooltip from going off-screen
  • Added custom scrollbar styling for better visual experience
  • Fixed tooltip disappearing immediately when trying to scroll

Improved

• Memory Graph Layout Optimization: Better handling of large node counts
  • Dynamic link distance based on node importance (80-180px range)
  • Stronger repulsion force for important nodes (-300 to -500)
• Adaptive label spacing based on node importance
  • Zone attraction strength decreases as node count increases
  • Larger radial distance for zones with many nodes
  • Significantly reduced node overlap in dense graphs

Fixed

• Critical: Fixed JavaScript syntax error in generated HTML
  • Removed duplicate }, 500); in setTimeout closure
  • HTML files now load correctly in all browsers
• Tooltip Interaction: Fixed tooltip disappearing when mouse moves to it
  • Added isMouseOverTooltip flag to track tooltip hover state
  • Tooltip stays visible when mouse transitions from node to tooltip
  • Only hides after mouse leaves both node and tooltip areas

Technical Details

• Added currentHoverNode variable to track active node hover
• Added isMouseOverTooltip flag for tooltip interaction state
• Enhanced collision detection with importance-based spacing
• Dynamic force parameters based on zone node density
• Improved tooltip positioning algorithm with boundary detection

Version Skip Declaration

• Due to a push error in 1.18.3, the version number was pushed to 1.18.4

v1.18.2 prevent destructuring errors when arguments are undefined

Fixed

• Fixed critical bug where undefined arguments caused "Cannot destructure property" errors
• Added safe argument handling for all Tool and Prompt calls
• Improved error messages when required parameters are missing

v1.18.1 enhanced tool descriptions and documentation

Improved

• Enhanced tool descriptions for better AI model understanding
• Added comprehensive multi-file context usage examples
• Improved documentation with v1.18.0 feature guidelines
• Optimized tool parameter descriptions for clarity

v1.18.0 Multi-file context support, enhanced search, and intelligent language detection

Added

• Multi-file Context Support: Record and track contexts spanning multiple files
  • New context_files table for storing file associations
  • Support for tracking change types, line ranges, and diff statistics per file
  • Automatic migration of existing single-file contexts to new system
  • Backward compatible with single file_path parameter
• Enhanced Search Performance: Multi-dimensional scoring and intelligent caching
  • Hybrid scoring combining semantic similarity (40%), keyword matching (30%), quality (20%), and freshness (10%)
  • LRU cache with 5-minute TTL for repeated searches
  • File-based filtering with new file_path parameter in semantic_search
  • Search result caching with use_cache parameter (default: true)
• Parallel Embedding Generation: Significantly faster vector embedding processing
  • Configurable concurrency with concurrency parameter (default: 5)
  • Batch processing with automatic error handling
  • Performance statistics including processing speed (embeddings/sec)
  • Up to 5x speed improvement for large context sets
• Intelligent Language Detection: Automatic project language detection for localized auto-recording
  • Analyzes code comments to determine project language (Chinese/English)
  • Falls back to README detection if comments are insufficient
  • Daemon auto-recording messages now use detected language
  • Supports bilingual projects with smart detection

Changed

• record_context tool now supports files_changed array parameter for multi-file recording
• semantic_search tool enhanced with file_path and use_cache parameters
• generate_embeddings tool now supports parallel processing with concurrency parameter
• Database schema extended with context_files table and indexes
• DatabaseManager.createContext() now automatically creates file associations when file_path is provided

Fixed

• Data migration now handles contexts with NULL file_path correctly
• File watcher initialization fixed for ES modules (replaced require with dynamic import)
• CLI and daemon now properly record file associations in context_files table
• CLI extract command now correctly records line number information for documentation files

Technical

• Added SearchCache class with LRU eviction and TTL support
• Added ContextFileManager for managing file-context relationships
• Added LanguageDetector utility for intelligent project language detection
• Implemented multi-dimensional scoring algorithm in VectorSearchEngine
• Added parallel embedding generation methods: batchGenerateEmbeddingsParallel() and batchUpdateEmbeddings()
• Database migration system for seamless upgrade from v1.17.0
• Lazy initialization pattern for ContextFileManager to avoid circular dependencies

v1.17.0 add CLI optimize, backup/restore commands and enhanced monitoring

Added

• CLI Optimize Command: Implemented memory optimization to clean up duplicate and low-quality contexts
  • devmind optimize <project-id> - Remove duplicates and low-quality contexts
  • --dry-run flag to preview changes without applying
  • Automatic detection of duplicate contexts (95%+ similarity)
  • Removal of low-quality contexts (score < 0.3, older than 60 days)
• CLI Backup/Restore Commands: Full database backup and restore functionality
  • devmind maintenance backup - Export database to JSON
  • devmind maintenance restore <file> - Import database from JSON backup
  • --output flag to specify custom backup location
  • --force flag to skip confirmation prompts
  • Automatic user confirmation before overwriting data
• Enhanced Git Monitoring: Improved daemon Git operation tracking
  • Real-time monitoring of .git/refs/heads directory
  • Automatic recording of commit messages, hashes, and changed files
  • Records author information and commit metadata
• Terminal Command Monitoring: Basic shell history tracking
  • Monitors .bash_history and .zsh_history files
  • Whitelist filtering (npm, git, node, python, docker, etc.)
  • Automatic sensitive information filtering (passwords, tokens, API keys)
  • --no-terminal flag to disable terminal monitoring
  • 5-second polling interval for new commands

Changed

• Database methods extended with optimization and export capabilities
• Daemon monitoring now includes both Git and terminal activity
• CLI commands now provide more detailed output and statistics

Technical

• Added findDuplicateContexts() method to DatabaseManager
• Added getLowQualityContexts() method to DatabaseManager
• Added deleteContextsBatch() method for efficient bulk deletion
• Added getAllProjects/Sessions/Contexts/Relationships() export methods
• Added clearAllData() method for database reset
• Enhanced daemon with configurable monitoring options

v1.16.2 Add type-to-zone mapping for memory graph visualization

Enhanced

• Memory Graph Type-to-Zone Mapping: New context types now automatically map to appropriate visualization zones
  • Added typeToZone mapping system to categorize 22+ context types into 6 visual regions
  • Code types (code_create, code_modify, code_refactor, etc.) → Blue Code/Test zone
  • Bug types (bug_fix, bug_report, bug_analysis) → Red Error zone
  • Feature types (feature_add, feature_update, feature_improvement) → Green Solution zone
  • Documentation types (docs_update) → Purple Documentation zone
  • Configuration types (dependency_update) → Pink Configuration zone
  • All 14 new fine-grained types properly cluster in their semantic zones

v1.16.1 Fix export_memory_graph schema and remove emoji decorators

Fixed

• Critical: Fixed export_memory_graph tool schema error that caused tool registration failure
  • Removed incorrect required: true from project_id property definition
  • Tool now correctly registers and can be called by AI assistants
• UI Cleanup: Removed emoji decorators from tool descriptions
  • update_quality_scores: Removed 🚀 [NEW] prefix
  • export_memory_graph: Removed 📊 [NEW] prefix
  • Improves consistency and reduces visual clutter in tool listings

Technical

• JSON Schema compliance: required field now only at inputSchema level
• Tool descriptions now follow consistent plain-text format

v1.16.0 Extended context types, enhanced metadata, and multi-file storage fix

Added

• 14 New Context Types:

  • Code: code_create, code_modify, code_delete, code_refactor, code_optimize
  • Bug: bug_fix, bug_report
  • Feature: feature_add, feature_update, feature_remove
  • Debug: debug_session, test_add, test_fix
  • Docs: docs_update

• Enhanced Metadata Fields:

  • Change tracking: change_type, change_reason, impact_level
  • Code analysis: affected_functions, affected_classes (auto-extracted)
  • Relations: related_files, related_issues, related_prs
  • Business: business_domain, priority
  • Stats: diff_stats, files_changed array

• Auto-Extraction:

  • Detects change types from code comments
  • Extracts function/class names (8 languages supported)
  • Assesses impact levels automatically
  • Parses issue/PR numbers from comments

Changed

• Memory graph now includes all 22 context types in filter dropdown
• Type filter organized into 6 groups: Core Types, Code Changes, Bug Related, Features, Debug & Test, Documentation
• Added Chinese translations for all new types
• Color mapping: new types inherit base type colors (blue/red/green/purple)

Fixed

• Multi-file Context Storage: Fixed file_path field incorrectly storing single file path for multi-file changes
  • When using files_changed array, file_path, line_start, and line_end are now correctly set to null
  • Actual file information properly stored in metadata.files_changed
  • Enhanced response message now shows all files in multi-file changes with individual diff stats
  • Fixes issue where multi-file commits appeared to only touch one file

Technical

• New EnhancedContextMetadata interface
• Enhanced ContentExtractor module with 7 analysis methods
• Zero schema changes (backward compatible)
• Multi-file tracking via files_changed array
• Auto-aggregation of diff stats across all files
• Auto-population of related_files from files_changed