arborist.nvim

WASM-first tree-sitter parser manager for Neovim 0.12+. Parsers install automatically when you open files. No manual steps, no maintenance.

Install

vim.pack.add({
  "https://github.com/arborist-ts/arborist.nvim",
})
require("arborist").setup()

Call setup() in your config — with or without options. To customize:

require("arborist").setup({
  update_cadence = "weekly",
  overrides = {
    my_language = { url = "https://github.com/me/tree-sitter-my-language" },
  },
})

All options and their defaults:

Option	Default	Description
prefer_wasm	true	Try WASM before native compilation
update_cadence	"daily"	"daily", "weekly", or "manual"
compiler	"cc"	C compiler for native .so builds
install_popular	true	Install popular language parsers at startup
ensure_installed	{}	Additional parsers to install eagerly at startup
ignore	{}	Extra filetypes to skip (merged with registry defaults)
overrides	{}	Extra parsers not in the registry

Parsers and Queries

Arborist takes a two-pronged approach so everything works out of the box:

Queries for 330 languages are bundled with arborist.nvim, providing syntax highlighting, folds, indents, and injections for every language Neovim supports — no parser needed yet.

Parsers for common languages are installed eagerly at startup when install_popular is enabled (the default). This covers the most popular programming languages, common config/data formats (JSON, YAML, TOML, XML, INI, Dockerfile, Makefile), and parsers needed by popular plugins like render-markdown.nvim (markdown, markdown_inline, html, latex). Set install_popular = false to disable. Use ensure_installed to add parsers beyond the popular set.

Everything else installs on demand — open a file and arborist handles the rest.

How It Works

1. You open a file
2. Arborist checks if a parser exists for that filetype
3. If not, it clones the source and builds one:
   • tree-sitter build --wasm (sandboxed, preferred)
   • tree-sitter build (native .so)
   • cc compilation (fallback)
4. Highlighting and indentation activate immediately

WASM builds are verified at load time — if the parser fails to load, arborist falls through to native compilation automatically. WASM is skipped entirely if your Neovim build lacks wasmtime.

Parser locations are resolved from a bundled community registry of 327 parsers. Unknown parsers fall back to convention-based lookup in the tree-sitter-grammars and tree-sitter GitHub orgs.

Batch installs group parsers by repository — parsers sharing a repo (e.g. typescript + tsx) clone once and build sequentially, while different repos clone in parallel.

User queries in ~/.config/nvim/queries/ always take highest priority.

Data Sources

Arborist bundles data from two companion repos so it works offline and ships cohesive versions:

Data	Source	Bundled in
Parser registry (327 parsers)	arborist-ts/registry	registry/
Query files (330 languages)	arborist-ts/queries	queries/

Changes flow upstream first: updates are made in the source repos, then synced into arborist.nvim before each release via scripts/sync-upstream.sh. This keeps the upstream repos canonical and arborist.nvim releases self-contained.

Commands

Command	Description
:Arborist	Show installed parsers and status
:ArboristInstall {lang}	Install a parser manually
:ArboristUpdate	Check all parsers for updates
:ArboristClean	Remove all arborist-managed parsers
:checkhealth arborist	Verify setup, tools, and bundled data

Requirements

• Neovim 0.12+
• git
• tree-sitter CLI (for building from source)
• A C compiler (fallback when tree-sitter CLI is unavailable)
• For WASM: Neovim built with ENABLE_WASMTIME=ON

License

MIT