CLI Agents

Terminal & CLI Agents

AI coding agents that live in your terminal or command line.

60 tools · 60 with full reviews · 35 free or open source · Updated Oct 7, 2026

Claude Code Freemium

Anthropic Claude Code is a terminal-native agentic coding tool that integrates directly into your development environment to plan, edit, and test code across entire repositories. It operates as an autonomous agent that can handle multi-file refactors, bug fixes, and feature implementations by reading your codebase and executing shell commands with your approval.

Codex CLI Freemium

OpenAI Codex CLI is an open-source, terminal-native coding agent powered by OpenAI's model family, including GPT-6 Sol and Luna. It enables developers to perform autonomous coding tasks, file edits, and shell command execution within a secure, sandboxed environment.

Sourcegraph Amp is an agentic coding tool that routes tasks across multiple frontier models and spawns parallel subagents to handle complex, multi-file engineering work. It operates across CLI, web, and IDE surfaces, focusing on large-scale repository context and team collaboration through shared threads.

Aider Aider is a terminal-based AI pair programmer that integrates directly into your local Git repository to edit files and auto-commit changes. It is model-agnostic, allowing users to connect to a wide range of frontier and local LLMs while maintaining full control over their coding workflow and costs.

Gemini CLI Paid

Google Gemini CLI is an open-source AI agent designed to bring Gemini model capabilities directly into the terminal for coding, content creation, and task automation. As of June 18, 2026, the tool has been deprecated for individual and free-tier users in favor of Google's new Antigravity CLI platform.

Block Goose is a general-purpose, open-source AI agent framework designed for code, workflows, and broad task automation. It runs locally on your machine via a desktop app, CLI, or API and is model-agnostic, allowing users to connect to over 25 LLM providers and extend functionality through the Model Context Protocol (MCP).

Amazon Q CLI Freemium

AWS Amazon Q Developer CLI is an agentic, terminal-based AI assistant designed to enhance developer workflows with deep AWS integration. It provides IDE-style autocompletion, natural language-to-bash translation, and the ability to perform multi-step tasks like file editing and infrastructure management.

Letta Code Freemium

Letta Letta Code is a memory-first, stateful agent harness designed for coding and autonomous tasks that persist and improve over time. It enables agents to maintain long-term memory, learn from experience, and adapt their own behaviors through research-backed techniques like sleep-time compute.

Qwen CLI Freemium

Alibaba Qwen Code is an open-source, terminal-first AI coding agent optimized for Qwen-Coder models that supports codebase exploration, automated file editing, and shell command execution. It functions as a flexible agentic platform that integrates with IDEs, supports multi-protocol API providers, and offers headless modes for CI/CD automation.

Codebuff Freemium

Codebuff Codebuff is a terminal-native AI coding agent that uses a multi-agent architecture to understand entire codebases, plan complex tasks, and execute precise file edits. It operates directly in the command line, allowing developers to perform structural refactoring, run tests, and install packages without leaving their terminal environment.

OpenClaw Open Source

OpenClaw OpenClaw is an open-source, vendor-neutral AI agentic framework designed for persistent, 24/7 operation across a user's own infrastructure. It functions as a personal operating system that manages tasks, communications, and system-level actions while keeping data and context local to the user's machine.

Command Code Freemium

Command Code Command Code is a coding agent that features a neuro-symbolic AI layer called taste-1, which continuously learns your specific coding preferences and patterns to provide personalized assistance. It supports a wide range of open-source and premium models, allowing users to switch between them mid-session while maintaining context and taste-based steering.

Charmbracelet Crush is a terminal-based AI coding agent that integrates directly into your development workflow by leveraging your existing tools, code, and terminal environment. It features a dual-agent architecture that routes tasks between large and small models to optimize for both performance and cost.

iFlow CLI Freemium

iFlow iFlow CLI is a terminal-based AI assistant that automates development tasks, repository analysis, and complex workflows through natural language interaction. It features a plugin-based architecture supporting SubAgents and Model Context Protocol (MCP) servers to extend its capabilities.

Kiro CLI Freemium

AWS Kiro CLI is an AI-powered terminal assistant that enables developers to build, test, and deploy applications through natural language commands. It integrates with your codebase to provide context-aware assistance, automate workflows, and manage cloud-based development sessions.

lucinate-ai Lucinate is a terminal-native chat client designed for distraction-free interaction with AI agents. It provides a keyboard-driven interface that supports multiple backends, including OpenClaw, Hermes, and various OpenAI-compatible endpoints like Ollama.

MCPJam Freemium

MCPJam MCPJam is a comprehensive testing and evaluation platform designed for developers building Model Context Protocol (MCP) servers. It provides tools to inspect, debug, and automate reliability testing for MCP servers across various AI clients like ChatGPT, Claude, and Cursor.

Mistral Vibe Freemium

Mistral Mistral Vibe (formerly Le Chat) is a unified AI agent platform designed for professional productivity and software development. It integrates chat, multi-step work automation, and autonomous coding agents into a single interface accessible via web, mobile, IDEs, and terminal.

Mux Open Source

Coder Mux is a desktop and web-based coding agent multiplexer designed for parallel agentic development. It allows developers to run multiple AI agents simultaneously in isolated workspaces, providing a platform for complex, long-running tasks that exceed the capabilities of traditional IDE-bound agents.

Zencoder Freemium

Zencoder Zencoder is an AI orchestration platform that provides autonomous agents for coding, testing, and business operations. It integrates directly into IDEs like VS Code and JetBrains, as well as external tools like Jira, Slack, and Notion, to automate multi-step workflows across the entire software development lifecycle and beyond.

Neovate Neovate Code is an open-source, feature-rich AI coding agent designed to enhance development workflows through natural language interaction. It supports a wide range of AI providers and models, offering capabilities like multi-file editing, spec-driven development, and a plugin system for custom workflows.

Sylph AdaL is an agentic coding tool designed to prioritize engineering quality and human review velocity over raw code generation speed. It functions as a terminal-first control plane that integrates with a browser-based agentic IDE to help developers plan, inspect, and ship code with confidence.

Cline Kanban Free

Cline Cline Kanban is a terminal-native, CLI-agnostic orchestration app designed to manage multiple AI coding agents in parallel. It provides a visual kanban board where each task card runs in an isolated git worktree, allowing developers to monitor progress, manage dependencies, and review diffs without juggling multiple terminal windows.

Parallel Code Free

Parallel Code Parallel Code is an open-source desktop application that acts as a workspace orchestrator for AI coding agents. It automatically manages git branches and worktrees to allow multiple AI agents to work on different tasks simultaneously without file collisions.

badlogic Pi is a minimalist, open-source coding agent for the terminal that provides a four-tool core (read, write, edit, bash) and a tree-structured interaction model. It is designed to be highly extensible via TypeScript, allowing developers to build custom workflows rather than relying on a pre-packaged, opinionated product.

Autohand ACP Freemium

Autohand Autohand ACP is an integration layer that enables the Autohand Code CLI to function as a native Agent Client Protocol (ACP) agent. It allows developers to embed Autohand's terminal-native coding capabilities directly into ACP-compatible IDEs like Zed with persistent session management.

Crow CLI Open Source

Crow Crow CLI is an Agent Communication Protocol (ACP) native coding agent that runs in your terminal or inside compatible editors like Zed. It features a streaming ReAct loop, multi-agent delegation, and robust session persistence that allows agents to recall past work and collaborate across sessions.

Fast Agent Open Source

EvalState Fast Agent is a CLI-first Python framework designed for building, testing, and orchestrating MCP-enabled agents and workflows. It provides a declarative syntax for composing multimodal agents that can interact with various LLM providers and external tools.

Minion Code Open Source

Femto Minion Code is an enhanced AI code assistant built on the Minion framework, designed as a specialized implementation of Claude Code. It provides a secure, pre-configured environment with robust tool integration for automated software development tasks.

Compass Nova CLI Open Source

Compass Nova is a terminal-first AI development assistant designed to understand your codebase, manage git workflows, and execute multi-step implementation plans. It provides a local-first experience that allows developers to perform complex coding tasks, refactoring, and debugging directly from the command line with granular safety controls.

Pi ACP Open Source

svkozak Pi ACP is an adapter that bridges the Agent Client Protocol (ACP) with the pi coding agent, enabling it to run within ACP-compatible editors like Zed. It manages session lifecycles, translates data between protocols, and supports pi-specific features like slash commands and session persistence.

Stakpak Agent Freemium

Stakpak Stakpak is a security-hardened, open-source DevOps AI agent that runs locally to automate infrastructure tasks, debug Kubernetes, and manage deployments. It operates as a 24/7 autonomous runtime that uses guardrails and secret substitution to perform operations without exposing sensitive credentials to LLMs.

OpenCode Freemium

Anomaly OpenCode is an open-source, provider-agnostic AI coding agent that operates via a terminal interface, desktop app, or web client. It supports over 75 model providers and features a client-server architecture where a background server manages sessions, tools, and configurations.

Antigravity CLI Freemium

Google Antigravity CLI is a terminal-based AI coding tool that provides a lightweight, keyboard-driven interface for interacting with Google's Antigravity agent harness. It enables developers to perform multi-file editing, bug tracking, and parallel task orchestration directly from their shell without context switching.

Nano Collective Nanocoder is a local-first, community-driven coding agent that runs directly in your terminal. It supports a wide range of AI providers, allowing users to bring their own models while maintaining strict privacy and control over their codebase.

AI Prompt Architect Freemium

AI Prompt Architect AI Prompt Architect is a professional prompt engineering platform that transforms vague user inputs into structured, deterministic, and production-ready prompts using the STCO (System, Task, Context, Output) framework. It provides an advanced engineering environment that supports multi-model comparisons, variable injection, and secure, encrypted prompt management for developers and enterprises.

PVR Labs AI Badger is a local-first codebase context extraction tool that prepares focused, token-efficient repository snippets for use with any AI chat interface. It operates entirely on your machine, ensuring privacy by requiring manual copy-paste actions to share context with external models.

AgentBox Open Source

Marco D'Alia AgentBox is a development orchestration platform that teleports local projects into isolated, ephemeral Linux environments to run multiple AI coding agents in parallel. It provides a secure, sandboxed computer experience for agents while keeping sensitive host credentials protected via a relay mechanism.

Oh My Pi Open Source

Can Bölük Oh My Pi (omp) is a terminal-native AI coding agent built on a high-performance Rust core that integrates IDE-grade tooling directly into the shell. It features hash-anchored edits for reliable patching, native LSP and DAP support, and robust subagent orchestration for complex coding tasks.

Grok Build Freemium

X Grok Build is a terminal-based AI coding agent that leverages parallel subagents and isolated Git worktrees to manage complex software engineering tasks. It supports a plan-and-approve workflow, allowing developers to review architectural changes before execution.

GCWing BitFun is a local-first, agentic desktop workspace that combines a high-performance Rust-based agent runtime with an integrated workbench for coding, office work, and personal automation. It supports multi-device control, custom agents, and interactive Mini Apps to turn tasks into persistent, visual workflows.

DeepSeek Harness Open Source

DeepSeek DeepSeek Harness is an open-source, plugin-based agent runtime that allows developers to compose custom coding agents by swapping components like models, tools, and sandboxes. Built on the Cordis framework, it treats every agent capability as a replaceable plugin, enabling high flexibility for research, evaluation, and custom agent development.

Atomic Agent Free

AtomicBot Atomic Agent is a local-first, open-source AI agent runtime that operates directly on your machine to browse the web, edit files, and execute shell commands. It is designed for privacy and performance, utilizing llama.cpp for local inference and maintaining all state, memory, and traces locally on your disk.

Kane CLI Freemium

TestMU AI Kane CLI is an AI-native browser automation tool that allows developers and AI agents to execute end-to-end web tests using plain-English objectives. It drives a real Chrome browser to perform tasks, self-heals when UI elements shift, and provides verified pass/fail results with shareable evidence.

Caveman Code Free

Julius Brussee Caveman Code is a terminal-based coding agent designed to minimize token consumption through aggressive, multi-layer context compression. It functions as a complete coding harness that optimizes prompts, tool outputs, and file reads to reduce costs in long-running agent sessions.

Sillage Open Source

MarlBurroW (open source) Sillage refers to two distinct open-source projects: a gradient-free memory and test-time learning tool for frozen language models, and a self-hosted private journaling platform for AI-assisted reflection. The memory tool provides fixed-size associative memory for LMs without requiring gradients or fine-tuning, while the journaling platform offers a private space for daily notes, history, and AI-based Q&A.

MathCode Open Source

Math-AI MathCode is a terminal-based AI coding assistant that converts plain-language mathematical problems into Lean 4 theorems and attempts to construct machine-verified proofs. It features a persistent Lean REPL for rapid iterative feedback and maintains a library of reusable theorems and axioms to build a cumulative knowledge base.

fx.sh Open Source

Vercel fx is a tiny, native coding agent harness written in Zig, designed for research, CI sandboxes, and embeddability. It functions as a Unix-like CLI tool that provides a shell-like experience rather than a full-screen TUI, supporting both local and cloud-based model inference.

Gitlawb Zero is an open-source, terminal-based AI coding agent that runs locally and provides full ownership over your development workflow. It supports a wide range of model providers, maintains durable local session history, and enforces strict, permission-gated access to your filesystem and shell.

Headlong Open Source

Laude Institute Headlong is an open-source agent microharness built in Bash that enables persistent agency, where agents maintain a continuous, self-guided inner monologue rather than waiting for reactive prompts. It treats human inputs as observations within a shared, ongoing stream of consciousness, allowing agents to autonomously set priorities, conduct research, and perform tasks.

Overbrilliant OB-1 is an open-source, terminal-native coding agent that automatically edits code and verifies changes by running project test suites. It features persistent memory, multi-agent orchestration, and a built-in model router that allows for immediate use without accounts or API keys.

Spotify Xirp is a macOS desktop application that serves as a vendor-neutral command center for managing multiple AI coding agent sessions. It allows developers to run agents in persistent terminal sessions across isolated Git worktrees, providing a unified interface for parallel development tasks.

Kolkrabbi Open Source

onembyte (Franco Michetti) Kolkrabbi is a model-agnostic, open-source AI coding agent for the terminal that allows users to switch between multiple AI subscriptions and local models within a single session. It features a unique effort-based scaling system and provides a local dashboard to track model performance and costs without telemetry.

FrontierAgent Free

Apodex AI FrontierAgent is an open-source agent runtime and terminal-based TUI designed for long-horizon research and file-based tasks. It supports two primary workflows: a stateful ReAct mode for sequential work and an Agent Team mode that uses a coordinator to manage parallel sub-agents.

YYLO Open Source

yylo-dev YYLO is an open-source platform that orchestrates AI coding agents by managing structured development workflows from initial intent to validated, production-ready code. It ensures complete traceability by preserving engineering memory, including task definitions, session details, and evaluation evidence within Git.

pi-agent-ide Free

Aleksandr Shpuntenko Pi Agent IDE is an agent-native extension for the Pi Coding Agent that provides a unified, semantic interface for reading, searching, and editing code. It equips the agent with specialized tools for terminals, debugging, and observability, allowing for more precise control over development workflows.

MiMo Code Freemium

Xiaomi MiMo Code is a terminal-native AI coding agent built as a fork of OpenCode that specializes in long-horizon tasks through persistent memory and autonomous subagent orchestration. It allows developers to manage complex, multi-step workflows directly from the shell while maintaining project context across sessions.

molt Open Source

Solvyx / Tyler Skelton molt (desktop app branded Maat Agent) is an open-source coding agent with a terminal CLI/TUI and an Electron desktop app. When the model says a task is done, molt runs the project's done.yml checks against the real files on disk and refuses the claim if any check fails, sending the failure back to the model.

Prime Agent Open Source

Prime Intellect Prime Agent is an open-source terminal coding and research agent from Prime Intellect built around Recursive Language Models. A persistent Python REPL is its built-in tool, so file operations, shell commands, context management and child subagents all run as code, while a daemon keeps sessions alive for long-running, autonomous work.

CodeAF Open Source

AgentField CodeAF is an open-source terminal coding harness from AgentField built to get frontier-grade results from open-weight models. It works as a "software factory" on your machine: one window for every project, requests split into parallel tasks on their own branches, and finished work merged only after it passes a check.

FAQ

What are the best CLI Agents tools?

The most complete CLI Agents listings on ai.dosa.dev include Claude Code, Codex CLI, Amp. Every tool has a detail page with pricing, key features, and a verdict - see the full list of 60 tools below.

Are CLI Agents tools free?

35 of the 60 CLI Agents tools listed here are free or open source. Others offer freemium or paid plans - check each tool's detail page for current pricing.

How is this list curated?

ai.dosa.dev is manually curated from community submissions and editorial review. Entries are enriched with pricing, features, and reviewer notes, and the directory is updated continuously.