Skip to content

NervaPack

Privacy-first, offline knowledge graph for developers

Build token-efficient context for your LLMs without sending code to the cloud.

Get Started View on GitHub

Why NervaPack?

NervaPack solves two fundamental problems with standard Vector RAG:

  • Token waste — chunk-based RAG retrieves blobs of text that may only tangentially relate to your query, bloating your context window.
  • Privacy risk — sending code to cloud embedding APIs leaks your proprietary logic.

NervaPack runs 100% on your machine. It uses tree-sitter to parse your codebase into a deterministic Abstract Syntax Tree graph, then uses a local Ollama model to draw hard semantic edges between your documentation and your code. Queries traverse this graph with a K-Hop BFS, returning a hyper-targeted, token-efficient context window — no cloud required.

Key Features

  • 100% Private

    All processing happens locally. Your code never leaves your machine. Uses ChromaDB + Ollama for complete privacy.

    Learn more

  • 91% Token Savings

    Graph-based retrieval reduces tokens by 91% vs naive RAG. Verified through real-world testing. Save on API costs and context window limits.

    Verified benchmarks · Token efficiency

  • :material-offline: Offline First

    Works completely offline with local Ollama models. Optional cloud providers (Claude, OpenAI) available.

    LLM providers

  • AST-Based Precision

    Deterministic parsing with tree-sitter. No arbitrary text chunks — only real code entities.

    How it works

  • Incremental Sync

    GitPython-powered surgical updates. Only changed files are re-indexed.

    Sync command

  • Rich Visualizations

    Interactive HTML graphs with community detection, search, and dependency analysis.

    Visualization guide

Quick Example

# Install (30 seconds)
brew install nervapack
# or: pipx install nervapack

# Build graph (2 minutes)
cd your-project/
nervapack ingest .

# Query (instant results)
nervapack query "How does authentication work?"

# Visualize
nervapack visualize --enhanced --communities

Output: Precise, token-efficient context with savings dashboard:

╭──────────────  NervaPack Token Efficiency  ──────────────╮
│  Strategy              Tokens   Visual            Relative │
│  Naive RAG (3 files)   12,840   ████████████████  100%    │
│  NervaPack              1,180   █░░░░░░░░░░░░░░░    9.2%  │
│ ──────────────────────────────────────────────────────────│
│  Tokens saved: 11,660   Reduction: 90.8%                  │
│  Cost saved (GPT-4o  $2.50/1M): $0.0292 per query         │
│  Cost saved (Claude Sonnet $3/1M): $0.0350 per query      │
╰───────────────────────────────────────────────────────────╯

Verified Performance

The 91% token reduction is verified through real-world testing on NervaPack's own codebase. See detailed benchmarks for test methodology and results.

NervaPack vs Standard Vector RAG

Standard Vector RAG NervaPack
Parsing Arbitrary text chunks Deterministic AST nodes (class, function, import)
Retrieval Nearest-neighbor blob K-Hop BFS on structural graph
Doc ↔ Code links None Hard EXPLAINS edges drawn by local LLM
Privacy Cloud embeddings 100% local (ChromaDB + Ollama)
Incremental sync Re-index everything Surgical per-file update via GitPython diff
Token savings No measurement Built-in dashboard shows exact reduction per query
Graph visibility Black box Interactive HTML visualization of every node and edge

Supported Languages

Bundled (no extra install): - Python, JavaScript, JSX, TypeScript, TSX

Optional extras: 

pip install "nervapack[go]"           # Go
pip install "nervapack[rust]"         # Rust
pip install "nervapack[java]"         # Java
pip install "nervapack[c]"            # C / C headers
pip install "nervapack[cpp]"          # C++
pip install "nervapack[ruby]"         # Ruby
pip install "nervapack[csharp]"       # C#
pip install "nervapack[all-languages]" # Everything

Use Cases

  • Code onboarding — Understand new codebases 10x faster
  • Documentation search — Find relevant docs linked to code
  • Refactoring analysis — See full dependency impact
  • LLM context optimization — 90% smaller prompts, same accuracy
  • Dependency auditing — Detect circular dependencies

What's Next?

Installation Guide Walk through setup for macOS, Linux, and Windows

Quick Start Tutorial Build your first knowledge graph in 5 minutes

Command Reference Detailed documentation for all 10 CLI commands

Community & Support

NervaPack is actively developed and maintained. We welcome contributions!

Contributing Guide