Ante: A Single-Binary Coding Agent That Runs Offline

Ante is a self-contained coding agent that lives in your terminal. It's a single ~15MB Rust binary from Antigma Labs with zero runtime dependencies. It works like Claude Code or Codex, but without their dependency chains or model constraints. You can run it fully offline with a local GGUF model.

Key Features

• Single binary: Hand-written Rust with Grep and git embedded — one process, no external tools.
• Natively offline: Uses a pinned, managed llama.cpp build. Point it at a GGUF file and the whole loop runs locally — no API key, no account, no internet.
• Low resource usage: Across 20 parallel Docker tasks, Ante uses ~7× less peak memory, ~9× less average CPU, and ~5× less disk I/O than Claude Code.
• Continuously benchmarked: Runs Terminal-Bench 2.1 under official leaderboard constraints. Latest full run: 82.7% with open-weight DeepSeek V4 Flash 0731 (368/445 trials, Ante 0.preview.71, ~$68 of inference).

Quick Install

Install with one command:

curl -fsSL https://ante.run/install.sh | bash

Install a specific release channel:

curl -fsSL https://ante.run/install.sh | bash -s -- nightly

Install to a directory already on PATH:

curl -fsSL https://ante.run/install.sh | ANTE_INSTALL_DIR=/usr/local/bin bash

Usage Modes

• Interactive TUI: ante — day-to-day work in the terminal
• Headless: ante -p "..." — one-shot tasks, scripts, CI
• Server: ante serve — editor plugins and integrations via JSONL protocol
• Gateway: ante gateway — run as Slack or Discord bot

Offline Mode Example

ante --offline-model ~/.ante/models/Qwen3.5-9B-Q4_K_M.gguf \
  -p "add error handling to src/main.rs"

Telemetry and Source

Telemetry is opt-out via ANTE_TELEMETRY=off. Sends only an anonymous installation label that you can re-mint. The core harness currently ships as a prebuilt binary; the repo holds docs, protocol, SDK, and eval pipeline — but they plan to address source release.