v0.67.0 · MIT License

Code with an
intelligent swarm.

CLI coding agent in Rust with native Agent Swarm for fast parallel requests, local-first models, and an embedded Code Knowledge Graph.

Install nowView source→
terminal
$wvc --provider ollama --model llama3.2 run 'build a REST API'
[swarm] Spawning 3 agents in parallel...
agent-1 Generating project structure
agent-2 Writing route handlers
agent-3 Setting up database schema
── 3/3 complete in 4.2s ──
Capabilities

Everything you need,
nothing you don't.

⬡

Agent Swarm

Fire multiple requests in parallel, orchestrated natively. No third-party orchestrator needed.

◉

Local Models First

Ollama, LM Studio, oMLX/llama.cpp, or any OpenAI-compatible endpoint. Your data stays on your machine.

◈

Code Knowledge Graph

Tree-sitter parses your codebase into a graph. "Who calls X?" "What depends on Y?" answered instantly.

⌕

Semantic Search

Local embeddings (all-MiniLM-L6-v2) find code by meaning, not just keywords.

↻

Incremental Indexing

Only re-indexes files that changed. Large codebases stay fast.

⊞

Cross-Platform

Linux, macOS, and Windows 11 (PowerShell 5.1+). One binary, three platforms.

wvccore binary
wvc-code-graphtree-sitter · SQLite · FTS5 · petgraph
wvc-embeddingall-MiniLM-L6-v2 · tract-onnx
wvc-swarm-coreagent orchestration · parallel dispatch
wvc-app-coretools · sessions · server
Get started

One command.
You're running.

macOS & Linux
curl -fsSL https://raw.githubusercontent.com/nicolasramos-es/weavecoder/main/install.sh | bash
Windows 11PowerShell 5.1+
irm https://raw.githubusercontent.com/nicolasramos-es/weavecoder/main/install.ps1 | iex
From source
git clone https://github.com/nicolasramos-es/weavecoder.git
cd weavecoder
cargo build --release --bin wvc
Quickstart

Three steps to your
first swarm.

1

Connect a local model

wvc login --provider ollama

Works with Ollama, LM Studio, oMLX, or any OpenAI-compatible endpoint.

2

Talk to the agent

wvc --provider ollama --model llama3.2 run 'hola'
3

Index your project

wvc init ./my-project --db ckg.db
wvc code-search "parseConfig" --db ckg.db

Tree-sitter extracts symbols, relations, and imports into SQLite + FTS5.

Ready to weave?

Open source. MIT licensed. Your code, your machine, your models.

Install WeavecoderStar on GitHub