download
One small binary. It filters your coding agent's terminal output locally, redacts secrets before they reach the model, and — only if you enroll it — reports numbers-only savings metrics to a dashboard.
Signed installers are coming. We won't ship an unsigned binary of a security product — until code-signing lands, build the agent from source below (2 commands), or contact us for an enterprise pilot with a managed rollout.
The engine is open-source, so the agent builds in a few minutes with a standard Rust toolchain — on macOS (Apple Silicon or Intel), Linux, and Windows.
# Prereq: Rust toolchain — https://rustup.rs
cargo build --release # inside the agent/ directory of the source tree
mkdir -p ~/.local/bin && cp target/release/ltk ~/.local/bin/
export PATH="$HOME/.local/bin:$PATH" # add to your shell profile too
ltk --version # -> ltk 0.43.0
Honest status: signed installers for every platform ship with the release pipeline; until then the agent builds from source in a few minutes.
Build from source today · signed & notarized installer next.
See instructions below · prebuilt + notarized builds are on the roadmap.
See instructions below · deb/rpm packaging scripts exist for enterprise fleets.
See instructions below · MSI packaging scripts exist; EV-signed installers are on the roadmap.
The agent is a standard Rust project (the toolchain is pinned in the repo). Any platform Rust supports:
git clone <the-lazytoken-repo>
cd <repo>/agent
cargo build --release
# binary lands at:
./target/release/ltk --version # -> ltk 0.43.0
# put it on your PATH, e.g.:
cp target/release/ltk ~/.local/bin/
Three steps: hook ltk into your coding agent, watch a command get filtered, and (optionally) connect a dashboard.
01 / hook in
Wires LazyToken into your AI coding agent's command path:
ltk init # Claude Code
ltk init --copilot # GitHub Copilot
ltk init --gemini # Gemini CLI
ltk init --codex # Codex CLI
For Cursor and other agents, run ltk init --show to see configuration options. --dry-run previews every change before writing.
02 / see it work
Run a command through ltk and compare with the raw output:
ltk git status
Same information, a fraction of the tokens. From now on your coding agent's git, test, docker, and 100+ other commands are filtered automatically.
03 / connect (optional)
Enterprise — your org's self-hosted server:
ltk enroll \
--server https://lt.company.internal \
--token ORG-XXXXXXXXXXXX
ltk status
Free (individual) — the hosted personal dashboard, once you have a sign-in token:
ltk enroll --cloud
Without enrollment, ltk filters locally and sends nothing anywhere.
Rolling out to dozens or hundreds of machines? Don't hand out tarballs — use the packaging + MDM scaffolding that ships with the product.
The repo's deploy/packaging/ tree builds native installers from a released binary:
LazyToken-<v>.msi), silent install with SERVERURL / ENROLLTOKEN properties.pkg with a post-install that enrolls as the console user.deb and .rpm (lazytoken-agent)Production signing (EV cert, notarization, GPG) is a release step documented in the packaging README — required before a regulated rollout.
Ready-made scripts and runbooks under deploy/mdm/, covered step-by-step in the Admin Guide:
.mst transform for parametersPre-seed the server URL and an org enrollment token, push through your MDM, and manage the fleet from the dashboard.
ltk update pulls from your own server's release channel and verifies sha256 + an Ed25519 signature before an atomic self-replace. You control the pace.
The 15-minute Quick Start takes you from zero to a live dashboard with demo data — Docker Compose or local Node, your choice.