gobby-code 0.2.6

The Problem

AI coding agents read entire files to find a single function. A 2000-line module gets dumped into the context window when all the agent needed was a 15-line method. Multiply that across a session and you're burning thousands of tokens on code that isn't relevant.

The Fix

gcode indexes your codebase using tree-sitter AST parsing and gives agents (and humans) precise, token-efficient access to symbols, search results, and dependency graphs.

$ gcode search "handleAuth"
[
  {"name": "handleAuth", "kind": "function", "file_path": "src/auth/middleware.ts",
   "line_start": 42, "signature": "async function handleAuth(req, res, next)", ...}
]

One search call instead of reading 50 files. 90%+ token savings.

How It Works

codebase → tree-sitter AST → SQLite index → search / retrieve / navigate
                │                   │
     ┌──────────┼──────────┐        │
     │          │          │        │
  symbols    chunks     files    ┌──┴──┐
  (FTS5)    (FTS5)   (hashes)   │     │
                              Neo4j  Qdrant
                             (calls) (vectors)

1. Index — Walk files, parse ASTs with tree-sitter, extract symbols and content chunks
2. Store — SQLite for symbols + FTS5, Neo4j for call/import graphs, Qdrant for semantic vectors
3. Search — Hybrid ranking: FTS5 + semantic similarity + graph relevance → Reciprocal Rank Fusion
4. Retrieve — Byte-offset reads for exact symbol source, no file-level bloat

Installation

Pre-built binaries

Download from GitHub Releases:

# macOS (Apple Silicon)
curl -L https://github.com/GobbyAI/gobby-code/releases/latest/download/gcode-aarch64-apple-darwin.tar.gz | tar xz
sudo mv gcode /usr/local/bin/

# macOS (Intel)
curl -L https://github.com/GobbyAI/gobby-code/releases/latest/download/gcode-x86_64-apple-darwin.tar.gz | tar xz
sudo mv gcode /usr/local/bin/

# Linux (x86_64)
curl -L https://github.com/GobbyAI/gobby-code/releases/latest/download/gcode-x86_64-unknown-linux-gnu.tar.gz | tar xz
sudo mv gcode /usr/local/bin/

# Linux (ARM64)
curl -L https://github.com/GobbyAI/gobby-code/releases/latest/download/gcode-aarch64-unknown-linux-gnu.tar.gz | tar xz
sudo mv gcode /usr/local/bin/

# Windows (x86_64) — PowerShell
Invoke-WebRequest -Uri https://github.com/GobbyAI/gobby-code/releases/latest/download/gcode-x86_64-pc-windows-msvc.zip -OutFile gcode.zip
Expand-Archive gcode.zip -DestinationPath .

Build from source

# With embeddings (requires cmake for llama-cpp-2)
cargo install --git https://github.com/GobbyAI/gobby-code

# Without embeddings (no cmake needed)
cargo install --git https://github.com/GobbyAI/gobby-code --no-default-features

With Gobby

gcode is installed automatically as part of the Gobby platform. If you're using Gobby, you already have it.

Usage

# Initialize and index a project (one step)
gcode init

# Search
gcode search "query"                      # Hybrid: FTS + semantic + graph boost
gcode search "query" --kind function      # Filter by symbol kind
gcode search-text "query"                 # FTS5 on symbol names/signatures
gcode search-content "query"              # FTS5 on file content

# Symbol retrieval
gcode outline src/auth.ts                 # Hierarchical symbol tree
gcode symbol <id>                         # Source code by symbol ID
gcode symbols <id1> <id2> ...             # Batch retrieve
gcode tree                                # File tree with symbol counts

# Dependency graph (requires Neo4j)
gcode callers "handleAuth"                # Who calls this?
gcode usages "handleAuth"                 # All references (calls + imports)
gcode imports src/auth.ts                 # Import graph for a file
gcode blast-radius "handleAuth" --depth 3 # Transitive impact analysis

# Project management
gcode status                              # Index stats
gcode projects                            # List all indexed projects
gcode index                               # Re-index (incremental)
gcode invalidate                          # Clear index, force full re-index

# Cross-project queries
gcode search --project myapp "query"      # By project name
gcode search --project /path/to/app "q"   # By path

# Global flags
--format text|json                        # Output format (default: json)
--quiet                                   # Suppress warnings and progress

Standalone vs Gobby

gcode works out of the box with zero dependencies — just gcode init and search. But it's designed to unlock its full potential with Gobby.

Standalone

codebase → tree-sitter → SQLite
                          (symbols + FTS5)

Full indexing and text search. No external services needed.

With Gobby

codebase → tree-sitter → SQLite        → FTS5 search
                          Neo4j         → call graphs, blast radius, imports
                          Qdrant + GGUF → semantic vector search
                          Gobby daemon  → auto-indexing, LLM summaries,
                                          config, secrets, sessions, agents

Gobby adds graph queries, semantic search, and infrastructure that makes gcode better at its core job — not just more features bolted on.

Search quality improves. With Neo4j, gcode search blends FTS5 text matching with call-graph relevance. Symbols that are heavily referenced rank higher. With Qdrant, conceptual queries like "database connection pooling" find semantically similar code even when the exact words don't match.

Summaries appear. gcode summary <symbol_id> returns AI-generated explanations of what a symbol does. The Gobby daemon generates these; standalone always returns null.

Config and secrets are managed. Neo4j URLs, Qdrant API keys, and auth credentials are stored in the shared database and encrypted with Fernet. No env vars to juggle.

Indexing happens automatically. The Gobby daemon watches for file changes and re-indexes in the background. Standalone requires manual gcode index.

Get started with Gobby at github.com/GobbyAI/gobby.

Graceful Degradation

Language Support

gcode parses ASTs using tree-sitter with support for 18 languages:

Build Features

The embeddings Cargo feature (default: on) enables local GGUF embedding generation via llama-cpp-2. Requires cmake to build. macOS builds use Metal GPU acceleration.

cargo build --release                        # With embeddings
cargo build --release --no-default-features  # Without embeddings (no cmake)

Platform Support

Contributing

See CONTRIBUTING.md for details.

License

Apache 2.0 — Use it, fork it, build on it.