# agentty **An agent harness for your terminal.** Ready to type in 47 ms, 21 MB of memory while idle.
One small native binary. Use the Claude, ChatGPT or Copilot plan you already pay for, or any model, even local.

Release Stars Downloads License Discord IRC

Website · Docs · Benchmarks · What is an agent harness? · Changelog

agentty fixing a failing test: plan, map the repo, run the tests, search, read, send a subagent, edit, patch, write a test, re-run, review
A real session with the current release, replayed from the terminal.

```bash curl -fsSL https://agentty.org/install.sh | sh # Linux · macOS irm https://agentty.org/install.ps1 | iex # Windows (PowerShell) ``` ## What it is The model decides what to do; the **harness** is everything around it: the tools it can call, the sandbox they run in, what goes into its context, which model handles which turn, and the interface you steer it from. agentty is that harness, written in C++26 and shipped as a single static binary. - **One small binary.** About 23 MB, starts in milliseconds, no Node, Python or Electron, no `npm install`. Download it and run it. - **Any model.** Sign in with the Claude Pro/Max, ChatGPT, GitHub Copilot or Kimi plan you already have, use an API key for OpenAI, DeepSeek, Gemini, Grok, Mistral, Groq, OpenRouter and more, or run a local model through Ollama. Switch mid-thread with `^P` and `^/`. - **Sandboxed by default.** Every shell command runs behind a wall the operating system enforces; your `~/.ssh`, `~/.aws` and `.env` files are masked. - **Frugal with context.** A local retrieval engine finds the code that matters instead of pasting whole files, so turns stay small and cheap. - **Everywhere you work.** In the terminal, inside Zed or JetBrains over ACP, as an MCP server, in scripts and CI, or on an air-gapped machine over SSH. - **Open source (MIT).** Coming from another tool? [vs Claude Code](https://agentty.org/compare/agentty-vs-claude-code/) · [vs Codex CLI](https://agentty.org/compare/agentty-vs-codex-cli/) · [vs Gemini CLI](https://agentty.org/compare/agentty-vs-gemini-cli/) · [vs opencode](https://agentty.org/compare/agentty-vs-opencode/) · [vs jcode](https://agentty.org/compare/agentty-vs-jcode/) · [vs Aider](https://agentty.org/compare/agentty-vs-aider/) · [vs Cursor](https://agentty.org/compare/agentty-vs-cursor/) · [all alternatives](https://agentty.org/alternatives/) ## Benchmarks What each harness costs you before the model does any work. Same machine, fresh install, no login, runs interleaved across tools; median of 10 launches (30 for `--version`). | | ready to type | idle memory | on disk, with runtime | `--version` | |---|---:|---:|---:|---:| | **agentty** 0.9.51 | **47 ms** | **21 MB** | **23 MB** | **2.6 ms** | | jcode 0.99.1 | 56 ms | 106 MB | 163 MB | 17 ms | | Codex CLI 0.167.0 | 402 ms | 152 MB | 536 MB | 77 ms | | Claude Code 2.1.319 | 635 ms | 152 MB | 325 MB | 14 ms | | Aider 0.86.2 | 1.0 s | 74 MB | 653 MB | 1.1 s | | Gemini CLI 0.67.0 | 3.0 s | 441 MB | 197 MB | 2.6 s | | opencode 1.19.5 | 5.2 s | 651 MB | 409 MB | 0.8 s | Measured 2026-11-08 on an Intel i5-6200U (4 threads, 16 GB), Linux. "Ready to type" is the time until the first prompt is fully drawn, read from the rendered terminal. agentty and jcode launch within the same range; call that a tie. Method, ranges and the script to reproduce it: agentty.org/benchmarks. ## Getting started ```bash cd your-project agentty ``` The first launch asks how to connect: sign in with **Claude Pro/Max, ChatGPT, GitHub Copilot or Kimi** (no API key), paste an API key, or pick a local Ollama model. Then type a request and press `Enter`. Rerun the install line to update. [Quick start ↗](https://agentty.org/docs/quick-start/) · [other installs](#installation) ## Features
### Fast all the way through A prompt on screen in under 50 ms. Independent tool calls run in parallel, and read-only tools start while the model is still writing, so tool-heavy turns finish sooner. ### Any model, any plan Subscriptions you already pay for, API keys for ten more providers, any OpenAI-compatible endpoint, or local Ollama. One model picker across every provider you're signed into. [Providers ↓](#providers) ### Sandboxed by default Shell and build commands run inside **claybin**: seccomp, Landlock and cgroups on Linux, seatbelt on macOS, AppContainer and job objects on Windows. Credential files are masked, file tools refuse paths outside your workspace, and network goes through a proxy that refuses your LAN, `localhost` and cloud metadata. Turn it off with `--sandbox off`. [Sandboxing ↗](https://agentty.org/docs/sandboxing/) ### Air-gapped mode `agentty airgap user@host` runs a session on a machine with no internet; your laptop relays the traffic over SSH with TLS pinned end to end. ### A full tool suite read · edit · write · apply_patch · shell · grep · glob · outline · find_definition · repo_map · git · test · diagnostics · web · search_code · search_docs · task · todo · remember — each with a purpose-built widget in the UI. ### Learns your codebase A fully local retrieval engine: `search_code` over your repo, `search_docs` over your docs, skills and memory. Keyword search works out of the box; `ollama pull nomic-embed-text` adds meaning-based search. Agent Skills in Claude Code's format, and `remember` / `forget` across sessions. [Retrieval ↗](https://agentty.org/docs/retrieval/) ### Smart Mode A flagship model plans; cheaper models do the legwork. Effort scales to each turn, and the router learns your repo across sessions. Off is a strict no-op. [Smart Mode ↗](https://agentty.org/docs/smart-mode/) ### Scripts and CI `agentty run "fix the failing test"` runs one prompt with tools and prints the answer. `--json`, `--timeout`, `--events jsonl`, clear exit codes, and `--read-only --no-network` for untrusted code. [Docs ↗](https://agentty.org/docs/cli/#agentty-run)
## Providers Sign in with a subscription you already have: ```bash agentty login # → Claude Pro/Max · ChatGPT (Plus/Pro) · GitHub Copilot · Kimi ``` Or point it at any of these: | `--provider` | What | Key | |---|---|---| | `anthropic` | Claude (the default): API key or Pro/Max sign-in | `agentty login` | | `chatgpt` | Codex models with your ChatGPT plan | `agentty login` | | `copilot` | GitHub Copilot models | `agentty login` | | `kimi` | Kimi K2 models | `agentty login` | | `openai` | GPT and o-series | `OPENAI_API_KEY` | | `deepseek` | DeepSeek V4 (chat and reasoner) | `DEEPSEEK_API_KEY` | | `gemini` | Google Gemini | `GEMINI_API_KEY` | | `xai` | xAI Grok | `XAI_API_KEY` | | `mistral` | Mistral, Codestral, Magistral | `MISTRAL_API_KEY` | | `groq` | Llama, Mixtral, Kimi on Groq LPUs | `GROQ_API_KEY` | | `cerebras` | Wafer-scale inference | `CEREBRAS_API_KEY` | | `together` | Open models on together.ai | `TOGETHER_API_KEY` | | `fireworks` | Open models on fireworks.ai | `FIREWORKS_API_KEY` | | `openrouter` | Any model via openrouter.ai | `OPENROUTER_API_KEY` | | `ollama` | Local models on `localhost:11434` | none | | `host:port` | Any OpenAI-compatible server (llama.cpp, vLLM, LM Studio…) | optional | ```bash agentty --provider ollama -m qwen2.5-coder # local, no key, no cloud agentty --provider deepseek -m deepseek-v4-pro agentty --provider openrouter # pick any model in-app ``` `--provider` persists. Inside a thread, `^P` switches provider and `^/` switches model, with no restart. Behind a gateway, `--auth-header X-API-Key` sends the key under a custom header. Full reference: [Providers & Models](https://agentty.org/docs/providers/). ## Keys | Key | Action | Key | Action | |-----|--------|-----|--------| | `Enter` | Send / queue | `^K` | Command palette | | `Esc` | Cancel / reject | `^J` | Thread list | | `S-Tab` | Cycle permission profile | `^P` | Provider picker | | `Alt+Enter` | Newline | `^/` | Model picker | | `^B` | Loop: resend until off | `^N` | New thread | | `^G` | Run a code block | `^S` | Smart Mode | | `^R` | Review changes | `^T` | Plan / todo | | `^O` | Tool outputs | `^U` | Read thinking (empty composer) | | `^L` | Stats | `^E` | Expand composer | | `^C` | Quit | `Alt+←/→` | Cycle threads | The composer is a readline-style editor (`^W`, `Alt+D`, `^Z`, `/` `@` `#` pickers, image paste). [Full keymap ↗](https://agentty.org/docs/keybindings/) ## Installation ```bash curl -fsSL https://agentty.org/install.sh | sh # Linux, macOS (builds from source if the binary won't run) brew tap 1ay1/tap && brew install agentty # macOS winget install agentty # Windows (or: irm https://agentty.org/install.ps1 | iex) yay -S agentty-bin # Arch ``` Every release is a single static executable for Linux (x86_64, aarch64), macOS and Windows.
.deb, .rpm, Scoop, Termux ```bash # Debian / Ubuntu curl -fsSLO https://github.com/agentty-org/agentty/releases/latest/download/agentty_${V}_amd64.deb sudo dpkg -i agentty_${V}_amd64.deb # Fedora / RHEL / CentOS sudo dnf install https://github.com/agentty-org/agentty/releases/latest/download/agentty-${V}-1.x86_64.rpm # openSUSE sudo zypper install https://github.com/agentty-org/agentty/releases/latest/download/agentty-${V}-1.x86_64.rpm # Windows (Scoop) scoop bucket add 1ay1 https://github.com/1ay1/scoop-bucket; scoop install agentty # Termux / Android (no root; commands run unsandboxed there) pkg install git cmake clang openssl libnghttp2 curl -fsSL https://agentty.org/install.sh | sh -s -- --build ``` `V` is the version, e.g. `0.9.53`. [Installation ↗](https://agentty.org/docs/installation/)
## Building from source Needs a C++26 toolchain (GCC 14+, a recent Clang, or MSVC). ```bash git clone https://github.com/agentty-org/agentty.git cd agentty && scripts/build.sh # debug preset, -j12; or: cmake --preset debug && cmake --build build -j 12 ``` Configure fetches missing submodules itself. `scripts/build.sh` builds at `-j12` by default (`AGENTTY_BUILD_JOBS=N` changes it), and Ninja pools sized from your RAM as well as your cores keep a full build from swapping your machine to death. The script also locks the build tree (a second build waits instead of corrupting the first) and repairs ninja logs left torn by an interrupted build. Use `-DAGENTTY_COMPILE_JOBS=N` / `-DAGENTTY_LINK_JOBS=N` to resize the pools. Editing the code? `cmake --preset debug` rebuilds in about 1.7 s after a one-line change (no LTO). [Building ↗](https://agentty.org/docs/building/#the-development-loop) ## More - **Air-gapped hosts:** `agentty airgap --setup user@host` once, then `agentty airgap user@host`. [Docs ↗](https://agentty.org/docs/airgap/) - **Zed, JetBrains (ACP):** `agentty acp`. Zed setting below. [Docs ↗](https://agentty.org/docs/acp/) - **MCP, both ways:** `agentty mcp-serve` exposes the native tools; servers in `.agentty/mcp.json` join them; `agentty mcp import` copies yours from another tool. [Docs ↗](https://agentty.org/docs/mcp/) - **Loop a prompt (`^B`):** resends after every turn until you press `^B` again, with backoff on errors and a `⟳ LOOP ×N` chip so it never looks idle. - **Run code blocks (`^G`):** runs a block from the last reply on your real terminal, then lets you attach the output. [Details](docs/RUN_CODE_BLOCK.md) - **Agent Skills:** drop a `SKILL.md` in `.agentty/skills/` (Claude Code's format works). [Installing skills](docs/website/skill-install.md) - **Architecture:** `(Model, Msg) -> (Model, Cmd)`, rendered by [maya](https://github.com/agentty-org/maya). [`docs/ARCHITECTURE.md`](docs/ARCHITECTURE.md) ```json { "agent_servers": { "agentty": { "command": "agentty", "args": ["acp"] } } } ``` ## Community [**Discord**](https://discord.gg/qhb9AZ8f3c) (with a helper bot that answers using agentty itself), or [**#agentty on Libera.Chat**](https://web.libera.chat/#agentty) / `#agentty:libera.chat` on Matrix. ## License MIT; see [LICENSE](LICENSE).