Skip to content

Changelog

All notable changes to NervaPack will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.4.1] - 2026-06-18

Fixed

  • Resolved network initialization race condition in visualizer
  • Fixed MCP server configuration paths

Changed

  • Added .nervapackignore for better file filtering
  • Updated MCP config with proper paths

[0.4.0] - 2026-06-16

Added

  • Multi-LLM provider support
    • Ollama (local, privacy-first)
    • Claude API (cloud, high quality)
    • OpenAI API (cloud, widely available)
    • MCP Delegation (zero-config for Claude Code/Cursor)
  • --llm and --model CLI flags for provider selection
  • Smart provider auto-detection (MCP → env vars → Ollama)
  • Cost estimation for cloud providers (shown before binding)
  • User confirmation prompts for cloud LLM usage
  • Provider validation and helpful error messages

Changed

  • Refactored LLM architecture to factory pattern
  • Split LLMSummarizer into provider-specific modules
  • Updated documentation with provider comparison tables
  • Enhanced CLI help text with provider examples

Documentation

  • Added comprehensive LLM provider guide
  • Created MULTI_LLM_IMPLEMENTATION.md with architecture details
  • Updated README with provider comparison and setup instructions

[0.3.1] - 2026-06-15

Fixed

  • Build and deployment configuration issues
  • Version bump for PyPI release

[0.3.0] - 2026-06-14

Added - Phase 3: Web Dashboard

  • Streamlit-based interactive web dashboard
    • nervapack serve command to launch on localhost
    • 4 dashboard tabs: Overview, Analytics, Query History, Graph Explorer
    • 8+ interactive Plotly charts (pie, bar, trends)
    • Caching strategy for fast subsequent loads (<200ms)
    • Real-time metrics with health score visualization

Added - Phase 2: Advanced Graph Visualization

  • Enhanced visualizer (v2) module
    • Real-time client-side search (instant filtering on 5000+ nodes)
    • Community detection using Louvain algorithm (10-color palette)
    • nervapack explore command for focused N-hop subgraph extraction
    • Interactive path finder with BFS shortest path algorithm
  • Dependency analysis module
    • nervapack dependencies command for file-level import analysis
    • Circular dependency detection
    • Hierarchical visualization with color-coded dependency types

Added - Phase 1: Enhanced CLI Visualizations

  • Graph analytics module
    • Health scoring algorithm (0-100 scale)
    • Language distribution metrics
    • Documentation coverage analysis
    • Connectivity and degree distribution stats
  • Query history module
    • Automatic query tracking to .nervapack/query_history.jsonl
    • nervapack history command with aggregate statistics
    • Cost savings dashboard across all queries
    • Most-queried topics analysis (word frequency)
  • Enhanced status command
    • --detailed flag for comprehensive analytics
    • Visual bars for language distribution
    • Top-10 most connected files
    • Git sync warnings

Changed

  • Updated query command with tree visualization of BFS traversal
  • Enhanced token savings panel with cost estimates
  • Improved visual presentation across all CLI commands

Documentation

  • Added VISUALIZATION_PLAN.md (5-phase roadmap)
  • Added phase completion docs (PHASE1_COMPLETE.md, PHASE2_COMPLETE.md, PHASE3_COMPLETE.md)
  • Added OVERALL_PROGRESS.md with cross-phase metrics
  • Updated KNOWLEDGE.md with comprehensive project overview

[0.2.0] - 2024-12-XX

Added

  • MCP server for Claude Code/Cursor integration
  • nervapack-mcp command-line tool
  • Three MCP tools: query_codebase, graph_status, list_entities
  • .mcp.json configuration support

Changed

  • Improved graph persistence and loading
  • Enhanced error messages

[0.1.0] - 2024-11-XX

Added

  • Initial release
  • Core commands: ingest, query, visualize, sync, status
  • Tree-sitter based AST parsing
  • NetworkX graph storage
  • ChromaDB vector store
  • Ollama LLM integration
  • GitPython incremental sync
  • PyVis interactive visualization
  • Token efficiency metrics

Supported Languages (Bundled)

  • Python
  • JavaScript
  • TypeScript
  • JSX
  • TSX

[Unreleased]

Planned Features

  • Additional language parsers (Go, Rust, Java)
  • Performance benchmarks
  • CI/CD examples
  • VS Code extension
  • Plugin system for custom parsers
  • Advanced query operators
  • Graph diff visualization

Version Numbering

NervaPack follows Semantic Versioning (SemVer):

  • MAJOR version (X.0.0): Incompatible API changes
  • MINOR version (0.X.0): New features (backward compatible)
  • PATCH version (0.0.X): Bug fixes (backward compatible)

Current Status

  • v0.4.x: Beta — Feature-complete, production-ready for early adopters
  • v1.0.0: Planned when test coverage >70% and docs are complete

Migration Guides

Upgrading to 0.4.0

Breaking changes: None

New features: - Multi-LLM provider support (no action required if using Ollama) - To use cloud providers: 

pip install "nervapack[claude]"  # For Claude API
pip install "nervapack[openai]"  # For OpenAI API

CLI changes: - New flags: --llm, --model, --api-key - Auto-detection works as before (no changes needed)

Upgrading to 0.3.0

Breaking changes: None

New features: - Web dashboard (requires optional install): 

pip install "nervapack[dashboard]"
- Enhanced visualizations work out of the box

Contributing

See Contributing Guide for development workflow and how to propose changes.

Report bugs and request features on GitHub Issues.