Rust·Latest release v0.3.8 · 2026-11-20

Rust Intelligence (ridx)

A Rust language server that answers from what rustc already worked out, read from an on-disk index.

Latest releasev0.3.82026-11-20
Last committoday2026-11-20
LicenseMIT OR Apache-2.0
Articles1on this blog
Recent releases
All 9 releases →
Recent commits
  • 4a9f235 release: name GitHub releases Rust Intelligence 2026-11-20
  • ef29a0a docs: cite the 0.3.8 Zenodo version 2026-11-20
  • 270a931 changelog: 0.3.8 Gatekeeper note for a browser-downloaded archive 2026-11-20
  • 70bab5a release: 0.3.8 2026-11-20

Status from GitHub as of 2026-11-20.

The problem

rust-analyzer keeps its own analysis of the whole workspace in memory and repeats name resolution, macro expansion and type inference that rustc performs on every cargo check. Rust Intelligence records what rustc resolved while cargo check compiles the workspace, through a rustc driver, into one binary shard per crate, and the language server answers editor requests from those memory-mapped shards.

Example uses

01. Editor language server

Use Rust Intelligence (the ridx binary) as the Rust language server in Zed (extension), VS Code, Neovim or Helix; with no subcommand it accepts rust-analyzer’s command line, so an editor set up for rust-analyzer can start it instead.

02. Shared dependency index

Index each dependency once per machine: shards go to a content-addressed store shared by every project and worktree, and std, core and alloc are indexed once per toolchain.

03. Queries from the shell

Ask for definitions, references, hover, types and symbols with ridx def, refs, hover, type-at and symbols, or compare an answer side by side with rust-analyzer using ridx compare.

Where it fits

Rust Intelligence does not analyse built code itself: a rustc driver writes the index during cargo check, and code typed since the last build is answered by a small compiler-free front end over that index. prod-code runs language servers and builds on a remote node; Rust Intelligence is a language server.

Get started

curl -fsSL https://rust-intelligence.com/install.sh | sh

The installer supports Linux x86_64 and arm64 and Apple Silicon macOS with rustup. It adds the Rust 1.98.1 toolchain that ridx-driver runs inside, downloads the release archive for the platform from GitHub, checks its SHA-256 and installs ridx and ridx-driver into ~/.local/bin. Since v0.3.8 the macOS binaries are signed with a Developer ID and notarized. v0.3.7 (2026-11-19) was the first public release; the earlier releases in the history predate it. Editor configurations are in the README and docs/editors/.

Authoritative installation steps, architecture diagrams, and full configuration references are maintained in the repository:
Read the Rust Intelligence (ridx) README on GitHub →

Official website →

Architecture →

Editor configurations →

Measured resources (raw results, Rust Intelligence f7105fcc) →

Limitations

  • ridx-driver links to the compiler’s private libraries and needs the Rust 1.98.1 toolchain; the installer adds it, and projects keep their own toolchain.
  • Answers come from the last build: code typed since then is answered from the open buffer, and a crate that fails to build keeps its last good index.
  • One cfg: like cargo check, the index sees one target and one feature set.
  • Linux x86_64 and arm64 and Apple Silicon macOS only; Intel macOS is not built and Windows is not supported.
  • The per-method agreement suite with rust-analyzer was not re-run for v0.3.7 or v0.3.8; the full 78-method matrix is for v0.2.0.
  • The published memory and CPU comparison is one run (n = 1) of Rust Intelligence f7105fcc on one Linux machine, and does not count the index build, which takes minutes on a large workspace the first time.
Cite this software · archived version v0.3.8

Cite this software

Alexander Panasenko (2026). Rust Intelligence (ridx) — a Rust language server answering from a compiler-built on-disk index (v0.3.8). Zenodo. https://doi.org/10.5281/zenodo.24387429

This record reflects the archived software version deposited on Zenodo for citation and reproducibility, which may differ from subsequent development releases.