Agentty is an open-source, high-performance native terminal-based AI coding agent written in C++ by 1ay1. Engineered specifically for speed and minimal resource consumption, Agentty combines an autonomous coding engine with a purpose-built terminal user interface subsystem. Operating directly in the terminal, Agentty handles interactive developer sessions, code modification, repository exploration, and tool calling. The application integrates dedicated C++ submodules including mcp-cpp for Model Context Protocol integrations, acp-cpp for agent communications, rag-cpp for codebase retrieval, and maya for low-latency terminal rendering. Day-to-day interactions involve continuous streaming code outputs, local file execution, and web research capabilities with built-in SSRF protections. Agentty differentiates itself primarily through its extreme focus on execution and rendering efficiency. Rather than relying on heavy runtimes like Node.js or Python, it runs as an optimized native binary featuring sub-millisecond per-frame TUI rendering, Myers bit-parallel grid diffing, hash-keyed measure caches, and bounded rendering overhead even across hundreds of turns and thousands of output lines. Agentty is open-source software available on GitHub under standard open-source workflows. It functions as a compiled native CLI tool with zero runtime language dependency overhead.
Native C++ Performance — Built in C++ with whole-program LTO and -O3 optimizations, delivering sub-3ms startup times and sub-millisecond per-frame TUI render latency. - Maya TUI Engine — Uses a custom terminal UI layer featuring Myers bit-parallel diffing, viewport paint clipping, blit-skipping, and hash-keyed layout caches for fluid output streaming. - MCP & ACP Integration — Natively incorporates Model Context Protocol (mcp-cpp) and agent communication protocols (acp-cpp) to interface cleanly with external tooling and model servers. - Embedded RAG — Integrates rag-cpp for local codebase indexing and context retrieval directly inside terminal workflows. - Hardened Web Tools — Provides native web search and retrieval tools with built-in DNS-rebinding and IP-resolution validation to prevent SSRF vulnerability. - Bounded Session Overhead — Employs frozen-prefix trimming to prevent per-frame render degradation during long multi-turn interactive coding runs.
Use cases
Use Case: Low-Latency Terminal Code Generation Developers working in resource-constrained environments or high-frequency terminal sessions use Agentty for fast, real-time code generation and editing without the latency and memory bloat of Electron or interpreted CLI runtimes.
Use Case: Context-Aware Codebase Exploration Engineers leverage Agentty's embedded rag-cpp and mcp-cpp submodules to query local codebases, index project structures, and inspect cross-file dependencies directly from the command line.
Use Case: Secure Network-Aware Autonomous Tasks Teams executing agents with web research requirements rely on Agentty's hardened web tooling to safely pull documentation and external APIs with post-resolution SSRF guards.
Who it is for
Agentty is built for systems engineers, C++ developers, and command-line purists who prioritize native performance, minimal latency, and low memory usage over heavy web-based UI tooling. It is less suitable for users looking for GUI/IDE plugins like VS Code extensions or those preferring browser-based workflows.
Install guide
Installation Agentty is compiled from source using CMake (3.28 or later) and a modern C++ compiler: bash git clone --recursive https://github.com/1ay1/agentty.git cd agentty cmake -B build -DCMAKE_BUILD_TYPE=Release cmake --build build -j$(nproc) sudo cmake --install build
First Run Configure required environment variables (such as your model API key) in your shell environment: bash export OPENAI_API_KEY="your_api_key_here" # Or configure relevant model endpoints supported via ACP/MCP agentty --help
Example Session Launch Agentty inside your repository to start an interactive terminal session: bash cd /path/to/project agentty "Analyze the CMake configuration and optimize test build flags"
GitHub Copilot CLI — GitHub Copilot CLI is a terminal-based AI coding assistant developed by GitHub. It integrates the Copilot coding agent runtime directly…
Aider — Aider is an open-source terminal-based AI pair programming tool created by Paul Gauthier and the Aider-AI community. It operates directly…
Terminai — Terminai (Termin.AI) is an open-source terminal process manager and AI agent harness developed by emosenkis. Built on top of the…
Amp — Amp (developed by AmpCode) is an autonomous coding agent platform and remote development harness designed to execute software engineering tasks…
Cline — Cline is an open-source autonomous coding agent runtime maintained by Cline Bot Inc. and an active open-source community. Available as…
Continue — Continue is an open-source AI code assistant and agent framework originally developed by the Continue team (continuedev) and recently acquired…