1
Anthropic's agentic coding tool for the terminal, IDE, desktop and web
★ 149k · +863 this week · updated 2026-10-01
9.4/10
What it is Claude Code is Anthropic's agentic coding tool. It reads your codebase, edits files, runs shell commands and tests, and works with git to create commits, branches and pull requests. The same engine runs on several surfaces: a terminal CLI, extensions for VS Code (and Cursor) and JetBrains IDEs, a Code tab in the Claude desktop app, and… Read more →
Pros
- Strongest ecosystem: skills, subagents, hooks, plugins and MCP
- Excellent at long, multi-step changes across a codebase
- Runs in terminal, IDEs, desktop and browser with one account
Cons
- No free plan; heavy use needs the more expensive Max plans
- AGENTS.md support has session-type exceptions (Bedrock, telemetry off) that still need a CLAUDE.md import
Pricing: Paid from $20/mo · terminal agent anthropic
🔥 Deal: Claude Pro billed annually: $17/month instead of $20
2
OpenAI's coding agent for the terminal, IDE and cloud, included with ChatGPT plans
★ 128k · +1.2k this week · Apache-2.0 · updated 2026-10-02
9.1/10
What it is Codex is OpenAI's coding agent. It comes in several forms: an open-source CLI (Apache-2.0) that runs locally in your terminal, IDE extensions for VS Code, Cursor and Windsurf, a desktop app launched with codex app, and Codex Web at chatgpt.com/codex, where tasks run in cloud sandboxes and can end in a pull request. You sign in with… Read more →
Pros
- Included with ChatGPT plans you may already pay for
- Open-source CLI plus cloud tasks that open pull requests
- Reads AGENTS.md, the cross-tool instructions file
Cons
- Fewer extension points than Claude Code (no plugin marketplace)
- Usage limits depend on your ChatGPT plan
Pricing: Free plan, paid from $8/mo · terminal agent openai
3
Open-source terminal coding agent that works with any model provider
★ 211k · +1.4k this week · MIT · updated 2026-10-02
8.4/10
What it is OpenCode is an open-source (MIT) AI coding agent built around a polished terminal interface, developed in the open at anomalyco/opencode. It is provider-neutral: you connect API keys from the LLM providers you already use, including local models, or use OpenCode Zen, the project's own gateway of tested models. A desktop app is available in beta for macOS,… Read more →
Pros
- Free and open source; any model provider
- Claude Code-style extensibility
- Fast, well-designed terminal UI
Cons
- You pay the model provider directly; costs vary with usage
- Smaller community of ready-made extensions
Pricing: Open source · terminal open source byok
4
Amp Freemium
Frontier coding agent for the terminal and editor, from the makers of Sourcegraph
8.2/10
What it is Amp is a coding agent and development environment from the team behind Sourcegraph, built around the latest frontier models. It runs as a terminal CLI, as a web app at ampcode.com, and as native macOS and iOS apps. Threads run either locally or on Orbs, cloud machines Amp creates per thread with your code, tools and plugins… Read more →
Pros
- Strong results on hard, multi-step tasks
- Shared threads for team review
- Subagents and AGENTS.md built in
Cons
- Usage-based costs add up on large tasks
- Fewer configuration knobs by design
Pricing: Free plan, paid from $20/mo · terminal agent
5
Orca Open source
An ADE for running fleets of coding agents in parallel worktrees, from desktop or mobile
★ 84k · +5.3k this week · MIT · updated 2026-10-02
8.0/10
What it does Orca is an application development environment (ADE) built around running several AI coding agents at once. Instead of driving one agent in one terminal, Orca spins up isolated git worktrees so Claude Code, Codex, OpenCode and other CLI agents can tackle the same prompt in parallel, and you compare their diffs side by side and merge the… Read more →
Pros
- Runs multiple coding agents in parallel isolated worktrees with diff comparison
- Free, open source (MIT), backed by Y Combinator
- Mobile companion apps for monitoring/steering agents remotely
Cons
- No hosted/managed tier documented — you bring your own agent subscriptions
- Extremely fast recent star growth is worth watching, even though the activity looks organic
Pricing: Open source · ade parallel-agents claude-code codex worktrees
6
Open-source AI coding agent from Alibaba's Qwen team, for terminal, IDE and desktop
★ 28k · +136 this week · Apache-2.0 · updated 2026-10-02
8.0/10
What it does Qwen Code is an open-source AI coding agent from Alibaba's QwenLM team that lives in your terminal and also ships as IDE plugins, a desktop app, and chat integrations. It is designed to be "agentic out of the box," handling multi-step coding tasks with memory and delegated sub-work rather than needing every step spelled out. What is… Read more →
Pros
- Official, well-funded project from Alibaba's Qwen team with transparent benchmark results
- Multi-protocol model support (OpenAI, Anthropic, Gemini, Qwen) avoids vendor lock-in
- First-class IDE plugins for VS Code, JetBrains and Zed, not just a terminal CLI
Cons
- "Agentic out of the box" features like Auto-Memory/Auto-Skills need review for what data they retain
- Best results likely come from pairing it with Qwen's own models rather than third-party ones
Pricing: Open source · qwen terminal-agent multi-model mcp jetbrains
7
Warp Freemium
Agentic development environment built on a modern terminal
★ 65k · +178 this week · AGPL-3.0 · updated 2026-10-02
7.9/10
What it is Warp began as a fast, Rust-based terminal and is now an agentic development environment: a modern terminal plus agents that write and refactor code, debug problems and run multi-step tasks while you stay in control. It runs on macOS, Linux and Windows. Warp's client is open source under AGPL v3. Beyond the desktop app there is a… Read more →
Pros
- Great terminal even without the AI
- Agents, code review and commands in one window
- Free plan to try it
Cons
- Credit-based AI usage on paid plans
- Account required
Pricing: Free plan, paid from $20/mo · terminal
🔥 Deal: Annual billing: Build plan $18/month instead of $20
8
Aider Open source
The original open-source AI pair programmer in your terminal
★ 49k · +149 this week · Apache-2.0 · updated 2026-05-22
7.8/10
What it is Aider is a long-running open-source (Apache-2.0) project for pair programming with a large language model in your terminal. You start it in a git repository, add the files you want to work on, and describe the change; Aider edits the files and commits each change to git with a descriptive message, so every step can be reviewed… Read more →
Pros
- Git-native: every change is a clean commit
- Works with nearly any model, including local ones
- Mature, well documented, very predictable
Cons
- No MCP servers, skills or subagents
- More hands-on than autonomous agents
Pricing: Open source · terminal open source git
9
Droids: coding agents for the terminal, IDE and CI from Factory
7.8/10
What it is Factory builds Droid, an AI agent platform for software development. The pitch is delegation: you hand Droid a task, review the diff, and merge, then scale from single sessions up to automated workflows across the whole development lifecycle. Droid runs in the Factory desktop app, the Droid CLI in your terminal, the web and mobile browsers, and… Read more →
Pros
- Works across terminal, IDE, Slack and CI
- Team and enterprise controls
- Custom droids for repeatable jobs
Cons
- Aimed more at teams than hobbyists
- Pricing based on usage tokens
Pricing: Paid from $20/mo · agent enterprise
🔥 Deal: Factory for Science: free Droid licenses for academic researchers, grad students, postdocs and research labs
10
Tuios Open source
A terminal window manager built for running fleets of AI coding agents
★ 4.6k · +836 this week · MIT · updated 2026-10-02
7.8/10
What it does Tuios ("Terminal UI Operating System") is a modern terminal multiplexer and window manager written in Go, built specifically around the workflow of running AI coding agents in the terminal. On top of a vim-modal tiling interface, it adds a layer purpose-built for agent supervision: it recognizes 20+ agent CLIs, tracks what each one is doing, and gives… Read more →
Pros
- Purpose-built agent state tracking and unified inbox across many CLIs
- Session persistence survives restarts and disconnects
- Actively maintained, MIT licensed, fast-growing (3.7k+ stars)
Cons
- Best experience needs a true-color, kitty-graphics-capable terminal
- Younger project than established multiplexers like tmux/zellij
Pricing: Open source · terminal ai-agents claude-code codex tmux-alternative
11
Goose Open source
Open-source, extensible AI agent, now stewarded by the Agentic AI Foundation
★ 55k · +229 this week · Apache-2.0 · updated 2026-10-02
7.6/10
What it is Goose is an open-source (Apache-2.0) AI agent that runs on your own machine as a desktop app for macOS, Linux and Windows, as a CLI, or through an API for embedding. It was created at Block and has since moved to the Agentic AI Foundation (AAIF) at the Linux Foundation for vendor-neutral governance; the code now lives… Read more →
Pros
- MCP-first design with a built-in extension browser
- Desktop app and CLI
- Any model provider, free and open source
Cons
- Less specialised for large refactors than coding-only agents
- Quality depends heavily on the model you choose
Pricing: Open source · agent open source mcp
12
Rust terminal coding agent that edits files, runs commands, and checks its own work
★ 41k · +4 this week · MIT · updated 2026-10-02
7.5/10
What it does Codewhale is an open-source terminal coding agent, written in Rust, that reads your project, edits files, runs commands, and checks its work using either a hosted model provider or a local model you choose. It positions itself as a fully open alternative to closed CLI coding agents, with the model backend left up to you rather than… Read more →
Pros
- Model-agnostic: works with hosted APIs or fully local models via Ollama/vLLM/SGLang
- Ships TUI, web UI and a VS Code extension from one core agent
- Genuinely active development history (10k+ commits) rather than inflated-looking stats
Cons
- Newer entrant competing in a crowded terminal-agent space with less brand recognition
- Full capability (multi-agent, computer use) requires careful approval-mode configuration
Pricing: Open source · rust terminal-agent local-models mcp vscode
13
Gptme Open source
Provider-agnostic terminal AI agent with shell, Python, file and browser tools
★ 4.4k · +12 this week · MIT · updated 2026-10-02
7.5/10
What it does gptme is an open-source, terminal-based personal AI agent. It gives an LLM direct access to your machine — a shell, a Python REPL, and file read/write/patch tools — so it can carry out multi-step coding and system tasks interactively, in a fully autonomous "agent" mode, or inside CI/CD pipelines. What is inside Out of the box gptme… Read more →
Pros
- Model-agnostic (Anthropic, OpenAI, Google, xAI, DeepSeek, OpenRouter, local llama.cpp) rather than locked to one vendor
- Mature and actively maintained: MIT, ~4.4k stars, 4,500+ commits, recent v0.33.0 release (Aug 2026) adding sandboxed execution
- Broad tool set out of the box: shell, Python REPL, incremental file patching, browser automation, vision, plus a plugin/skills/hooks system
Cons
- Executes arbitrary shell and Python commands on your machine at the model's direction — real risk if run unsupervised without its guardrail features enabled
- Autonomous/persistent-agent mode needs careful configuration (task selectors, pre-commit checks) to use safely
Pricing: Open source · terminal-agent cli python shell autonomous-agent
14
Self-hosted unified API for running Codex, Claude Code, and other agent harnesses
★ 2.8k · +31 this week · Apache-2.0 · updated 2026-10-02
7.5/10
What it does HarnessRouter Community Edition is a self-hosted infrastructure layer that puts a single, unified API in front of multiple AI coding-agent harnesses — Codex, Claude Code, Hermes, PI, DSH, and others. Instead of writing separate integration code for each agent CLI, you talk to one Gateway that implements the Unified Harness Protocol (UHP) and exposes an OpenAI Responses… Read more →
Pros
- One OpenAI-Responses-compatible API in front of multiple coding-agent harnesses, so products can swap backends without rewriting integration code
- Apache-2.0, self-hosted via a single Docker command, with session isolation, streaming, file handling, and secret redaction built in
Cons
- Ships with default admin credentials (harnessrouter/harnessrouter) that must be changed immediately; Community Edition is new enough that production track recor
Pricing: Open source · agent-api agent-harness self-hosted mcp docker
15
Herdr Open source
Persistent terminal runtime that hosts Claude Code, Codex, Cursor and other coding agents
★ 42k · +1.1k this week · Apache-2.0 · updated 2026-10-02
7.5/10
What it does Herdr is a terminal runtime — a tmux-like session host, not a coding agent itself — built specifically to run AI coding agents persistently. It keeps agent sessions alive in a background server so you can detach, close your laptop, or lose an SSH connection without killing the work, and gives you one window across your local… Read more →
Pros
- Single Rust binary, no Electron — a lightweight, cross-platform runtime rather than another agent wrapper
- Persists agent sessions across disconnects/restarts and unifies local plus remote SSH machines in one window
- Explicitly documented to work with Claude Code, Codex, Cursor, OpenCode, and Grok, with per-pane working/blocked/idle status
Cons
- It's a session host/multiplexer, not a coding agent itself — you still need one of the supported agents to actually write code
- It is genuinely a runtime/host for terminal agents rather than a coding agent in its own right, so "terminal-agents" is the closest fit among the given categori
Pricing: Open source · terminal rust session-persistence remote-ssh multiplexer
16
Zero-config macOS terminal built on WezTerm for running AI coding CLIs
★ 6.1k · +76 this week · MIT · updated 2026-09-28
7.5/10
What it does Kaku is a free, open-source macOS terminal emulator built on top of WezTerm, purpose-tuned for coding with AI CLI agents. Rather than a blank terminal you have to configure yourself, Kaku ships with pre-set fonts, themes, tab/pane management, and shell integration so it works well out of the box. What is inside Kaku bundles keyboard-driven shortcuts for… Read more →
Pros
- Popular and active: ~6k GitHub stars, 1,660+ commits, MIT licensed
- Ships with sensible zero-config defaults, fonts, themes and shell integration tuned for running AI coding CLIs
- Built-in glue for a real workflow: pane/tab management plus Lazygit and Yazi integration alongside AI chat/command-suggestion shortcuts
Cons
- macOS only
- It's a terminal host, not an agent itself — you still bring your own AI CLI (Claude Code, Gemini, Copilot, Kimi Code) and API keys
Pricing: Free · terminal macos wezterm ai-coding cli
17
Abtop Open source
htop-style live dashboard for your Claude Code, Codex and OpenCode sessions
★ 3.7k · +41 this week · MIT · updated 2026-09-14
7.0/10
What it does Abtop is a terminal dashboard, styled after htop, purpose-built to monitor AI coding agent sessions rather than system processes. Instead of CPU/RAM per process, it shows per-session token usage, context-window percentage, rate limits, git status, running child processes, open ports, and current task, refreshed live while you work. What is inside This isn't a tool the agent… Read more →
Pros
- Read-only, no API keys or auth required, per its own README
- Tracks token usage, context %, rate limits and git status live across multiple agent sessions
- 12 themes including colorblind-friendly options; MIT licensed with ~3.6k stars and active development
Cons
- It's a monitoring companion, not a coding agent itself, so it's a loose fit for an 'app' directory built around agents
- Codex CLI and especially OpenCode support are partial (missing context %, current task, rate limit, or requiring local sqlite3 access)
Pricing: Open source · monitoring terminal claude-code rust dashboard
18
CLI that gives AI coding agents shared, version-controlled project memory via MCP
★ 5.0k · +-5 this week · updated 2026-06-25 · archived
7.0/10
What it does ByteRover CLI (brv) gives AI coding agents persistent, structured memory. Run brv in a project and it opens an interactive REPL, powered by an LLM of your choice, that curates project knowledge into a "context tree," lets you query it, and syncs it to the cloud so the same knowledge can be shared across tools and teammates.… Read more →
Pros
- Deep integration matrix — 22+ coding agents, 20 LLM providers, and a native MCP server (`brv mcp`) for other agents to query
- Git-like version control for the context tree (branch/commit/merge/push/pull) plus a human-in-the-loop review workflow (`brv review pending/approve/reject`)
- Publishes concrete benchmark numbers (LoCoMo, LongMemEval-S) rather than only marketing claims
Cons
- Elastic License 2.0 is source-available, not fully open source — it restricts offering the software as a competing hosted service
- Its main value (ByteRover Cloud sync, hosted LLM, team management) pushes toward a paid tier beyond the free local CLI; last commit was ~3 months ago, slower-mo
Pricing: Freemium · memory context mcp repl multi-llm
19
Web workspace for Claude Code: chat, Kanban tasks, scheduling, multi-agent bots
★ 141 · +3 this week · updated 2026-09-28
7.0/10
What it does Claude Code Studio wraps Anthropic's Claude Code CLI in a self-hosted web (and desktop) dashboard, letting you drive AI-assisted development from a browser or phone instead of a bare terminal. It streams chat responses, lets you paste screenshots and attach files, and keeps session memory across conversations. What is inside - A Kanban board for queuing and… Read more →
Pros
- Full-featured dashboard for Claude Code with Kanban, scheduling and multi-agent bots
- Runs on desktop, web and Docker with no build step required
- Security-conscious defaults (bcrypt, AES-256-GCM, path-traversal protection)
Cons
- Small community (136 stars) with a single primary maintainer
- Self-hosted web server on top of Claude Code increases attack surface if exposed publicly
Pricing: Open source · claude-code dashboard kanban multi-agent scheduler
20
Open-source Rust terminal coding agent with ACP support for Zed, Neovim, JetBrains
★ 10k · +3 this week · GPL-3.0 · updated 2026-09-02
7.0/10
What it does Claurst is an open-source, multi-provider terminal coding agent written in Rust. It began life as a clean-room reimplementation of Claude Code's observed behavior — built from an independently produced specification, never the original source — and has grown into its own TUI pair-programmer with a companion mascot ("Rustle"), chat forking, memory consolidation and a plugin system. What… Read more →
Pros
- Speaks the open Agent Client Protocol, so it plugs into Zed, Neovim, and JetBrains as a native agent
- Multi-provider AI routing rather than lock-in to one vendor, plus a plugin system and chat forking
- Fully open source (GPL-3.0) with no telemetry/tracking by the maintainer's own account; actively developed (commits within the last month)
Cons
- Still officially Beta (v0.1.7); the README itself flags /share, Free Mode, /goal, and "ultracode" as experimental
- Originated as a clean-room reimplementation of Claude Code's behavior rather than an original design, so parity with the original may lag
Pricing: Open source · terminal rust acp multi-provider open-source
21
Cmux Freemium
Ghostty-based macOS terminal built for running many AI coding agents at once
★ 28k · +153 this week · updated 2026-10-02
7.0/10
What it does cmux is a native macOS terminal, built on libghostty, designed around running several AI coding agent sessions side by side rather than as an afterthought bolted onto a generic terminal. Vertical and horizontal tabs, per-pane notifications, and a sidebar showing git/PR/port state are all built for juggling many concurrent agents. What is inside - Blue notification rings… Read more →
Pros
- Purpose-built notification and status system (rings, tab highlights, a dedicated panel) so you can tell at a glance which of many agent panes needs attention
- Sidebar surfaces git branch, PR status, working directory, and listening ports per pane; session layout, scrollback, and browser history survive a relaunch
- Free and open source (GPL-3.0-or-later) with native resume integrations for a long list of CLI agents beyond Claude Code
Cons
- macOS only, and still early with heavy churn (2k open issues, 1.7k open PRs) relative to its 27k stars
- Founder's Edition (paid) gates the iOS app, cloud VMs and early AI features behind a tier with no public pricing listed
Pricing: Freemium · terminal claude-code codex opencode cli
22
DeepSeek-native terminal coding agent built for stable, long-running sessions
★ 36k · +28 this week · MIT · updated 2026-10-02
7.0/10
What it does DeepSeek Reasonix is a terminal AI coding agent distributed as a single Go binary, built around DeepSeek models but engineered to work with any OpenAI-compatible endpoint. Its headline design goal is prefix-cache stability, so long-running, always-on agent sessions stay cheap and consistent instead of losing cache and re-paying for context on every turn. What is inside -… Read more →
Pros
- Cache-stability focus specifically targets cost and reliability of long-running sessions
- One core agent exposed as CLI, desktop app, VS Code extension and browser client
- Signed Windows installers and documented security policy
Cons
- Optimized around DeepSeek models by default, requiring extra config for other providers
- Extension Protocol v1 is project-specific rather than a widely adopted standard
Pricing: Open source · deepseek terminal-agent golang mcp vscode
23
FuXi Freemium
Provider-agnostic terminal coding agent positioned as a bring-your-own-model Claude Code alternative
★ 3.3k · +-16 this week · updated 2026-09-29
7.0/10
What it does FuXi is a terminal-based, provider-agnostic AI coding agent — pitched as an alternative to Claude Code that works with any OpenAI-compatible model, plus Gemini and Bedrock/Vertex. It runs a "Think → Act → Verify" loop: reading and editing code, running shell commands, and verifying its own changes, with cost-aware routing and automatic failover across whichever providers you… Read more →
Pros
- Feature-rich single static binary: 50+ tools, LSP diagnostics, Jupyter, browser automation, MCP client, durable sessions with checkpoints
- Unusually transparent evaluation practice — publishes a reproducible, self-run benchmark against Claude Code with explicit caveats instead of a marketing score
Cons
- Source code is proprietary despite living on GitHub — the repo hosts only docs, installers and the issue tracker, not the implementation
- Installs via curl-pipe-to-shell, and ships a flag that bypasses all permission checks — treat it the way you would any command with root-level reach
Pricing: Freemium · terminal-agent provider-agnostic mcp-client proprietary go-binary
24
Open-source terminal coding agent for xAI's Grok API with search, media generation and remote control
★ 3.5k · +5 this week · MIT · updated 2026-07-06
7.0/10
What it does grok-cli is an open-source terminal coding agent built specifically around xAI's Grok API. Beyond the standard "read code, edit files, run commands" agent loop, it adds live X/web search, foreground and background sub-agent delegation, a built-in desktop-automation sub-agent, and remote control of a running session from Telegram. What is inside Foreground task and background delegate sub-agents, a… Read more →
Pros
- Fully open source (MIT) with a broad, well-documented feature set: sub-agents, MCP, lifecycle hooks, a microVM sandbox, and headless/CI-friendly modes
- Backed by a real org (superagent-ai) with CI, changelog and clear troubleshooting docs
Cons
- Main branch's tip commit is from May 2026, several months before this review, suggesting a slower recent cadence
- Desktop automation and Telegram remote control widen the attack surface; sandboxing is Apple-Silicon-only, so Linux/Intel Mac runs are unsandboxed
Pricing: Open source · grok xai terminal-agent open-source mcp
25
Hunk Open source
Terminal diff viewer built for reviewing multi-file changes made by AI coding agents
★ 9.5k · +69 this week · MIT · updated 2026-10-01
7.0/10
What it does Hunk is a terminal-based, review-first diff viewer purpose-built for looking at code changes produced by AI coding agents, rather than plain unified diff text. It renders a multi-file review stream with sidebar navigation, split/unified/auto layouts, and syntax highlighting, so a session where an agent touched a dozen files is easier to step through than scrolling raw terminal… Read more →
Pros
- Purpose-built review UI for agent-generated diffs: multi-file review stream, inline agent annotations, split/unified/auto layouts, watch mode
- Works as a Git pager/difftool and auto-detects Jujutsu and Sapling too, not just Git
- Actively maintained (MIT, 9.3k stars, 305 forks, 876 commits) with an extension system and community Discord
Cons
- Doesn't map cleanly onto any single existing app category here — it's a review/viewer tool for agent output, not an agent, editor, app-builder, or cloud agent i
- No dated "last commit" was visible in the fetched page, so exact recent-activity cadence couldn't be confirmed beyond the commit count
Pricing: Open source · diff-viewer code-review terminal-ui git-pager agent-review
26
Kun Freemium
Local-first AI agent workspace for code, design, and documents, in GUI or TUI
★ 6.3k · +3 this week · updated 2026-10-02
7.0/10
What it does Kun is a local-first AI agent workspace built to move from "AI answering questions" to "AI completing work." It runs as a single shared runtime behind either a desktop GUI or a terminal TUI, so tasks, approvals, plans and evidence carry over no matter which interface you open. What is inside - Code mode: project-aware file editing,… Read more →
Pros
- Single runtime spans desktop GUI and terminal TUI with shared state
- Broad scope beyond coding: design canvas and document/office modes too
- Multi-model support (Claude, ChatGPT, Gemini, Ollama, DeepSeek) avoids vendor lock-in
Cons
- PolyForm Noncommercial license blocks commercial use without a separate agreement
- Breadth across code/design/work modes may mean less depth than a coding-only specialist tool
Pricing: Freemium · local-first agent-workspace tui mcp multi-model
27
Terminal coding agent with real LSP refactors and a live debugger built in
★ 34k · +770 this week · MIT · updated 2026-10-02
7.0/10
What it does Oh My Pi (omp) is a terminal coding agent built by wiring real IDE functionality directly into the agent loop — its tagline is "a coding agent with the IDE wired in." Instead of just editing text, it uses actual Language Server Protocol (LSP) connections for workspace-aware renames/refactors and a real Debug Adapter Protocol (DAP) debugger (lldb,… Read more →
Pros
- Real LSP and DAP integration gives IDE-grade refactors and live debugging (lldb, dlv, debugpy), not just text-based edits
- Imports existing config from Claude Code, Cursor, Windsurf, Codex, and Cline, easing migration from those tools
- Broad tool surface: 60+ model providers, 31 built-in tools, web search across 23 providers, browser automation, and subagents
Cons
- Large attack/blast-radius surface (desktop control, code execution, browser automation) whose safety depends heavily on permission mode and provider choice
- Very fast-moving, code-heavy project (~80k-line Rust core, 24,000+ commits) — a lot of surface area for users to audit themselves
Pricing: Open source · terminal cli lsp debugger subagents
28
Terminal coding agent tuned for cheap open models like Kimi K3 and GLM 5.3
★ 68k · +43 this week · Apache-2.0 · updated 2026-10-02
7.0/10
What it does Open Interpreter is a terminal-based coding agent, rewritten in Rust as a fork of OpenAI's Codex harness, aimed at getting strong agentic performance out of budget-friendly open-weight models (the README calls out Kimi K3 and GLM 5.3) instead of only frontier proprietary ones. It executes commands in native sandboxes and can drive both terminal and GUI applications.… Read more →
Pros
- Optimized specifically for cheap/open-weight models, not just frontier APIs
- Large, active project with extensive documentation and multiple harness modes
- Supports MCP, skills, hooks and computer-use out of the box
Cons
- Major architectural departure from the original Python Open Interpreter concept may confuse users expecting that project
- Broad shell/computer-use capabilities require careful sandbox and permission configuration
Pricing: Open source · terminal-agent rust open-models codex computer-use
29
Local-first Electron/Rust desktop workspace for AI coding agents with subagent delegation
★ 6.3k · +523 this week · LGPL-3.0 · updated 2026-10-02
7.0/10
What it does PI Desktop is a local-first desktop application, built on Electron with a Rust host core and a "pi Agent Harness," that gives AI coding agents a persistent workspace instead of a disposable terminal session. Work is organized as a Project → Session → Agent hierarchy rather than a single throwaway chat, so context, decisions and files stay… Read more →
Pros
- Local-first with no mandatory relay service, keeping data on your machine
- Three distinct operating modes (Agent/Plan/Goal) give explicit control over autonomy level
- Real subagent delegation and session branching, not just a single linear chat
Cons
- Requires a Rust toolchain and specific Node/pnpm versions for building from source
- Moderate community size (4.5k stars) compared to some larger competitors in this list
Pricing: Open source · electron rust local-first mcp subagents
30
Multi-user, multi-agent real-time collaboration space for humans and coding agents
★ 3.8k · +10 this week · Apache-2.0 · updated 2026-09-05
7.0/10
What it does Tutti is a real-time shared workspace where people and AI coding agents work in the same connected room instead of copying files and context back and forth between separate agent sessions. Rather than replacing Claude Code or Codex, it sits alongside them as a collaboration layer both humans and agents can see and act in together. What… Read more →
Pros
- Addresses a real pain point: manual context-copying between separate agent tools
- Genuine open-source edition alongside an optional hosted multi-user version
- Active development (5,300+ commits) and a real, if moderate, community
Cons
- Multi-user cloud edition (Tutti VM) needs its own access-control review for sensitive code
- Value depends on already running multiple different coding agents side by side
Pricing: Freemium · multi-agent collaboration claude-code codex desktop
31
Local-first desktop workspace for Claude Code and other agents, with worktrees and diffs
★ 15k · +115 this week · MIT · updated 2026-10-01
7.0/10
What it does cc-haha is a local-first, cross-platform desktop application (Electron) that wraps Claude Code and other AI coding agents in a workspace UI, so you can run multiple agent sessions, review their changes, and manage git branches without living entirely in a terminal. What is inside - Multi-session workspaces with global search across past sessions - Git branch and… Read more →
Pros
- Strong review workflow: worktrees, per-file diffs, and an embedded browser preview
- Supports multiple model providers, not just Claude
- Large, genuinely active community (14.6k stars, 2,000+ commits, active issues/PRs)
Cons
- Many third-party chat integrations (Telegram, WeChat, etc.) widen the permissions/attack surface to review
- Unsigned installers on some platforms require manual security approval to run
Pricing: Open source · claude-code electron desktop git-worktrees diff-review
32
Google's open-source terminal agent, retired for consumer accounts in June 2026 (enterprise and API keys only)
★ 107k · +41 this week · Apache-2.0 · updated 2026-10-02
6.8/10
Status (September 2026): Google retired Gemini CLI for free users, Google AI Pro and Ultra subscribers and individual Gemini Code Assist users on 18 June 2026, and replaced it with the Antigravity CLI. It keeps working for Gemini Code Assist Standard and Enterprise licences, Gemini Code Assist for GitHub through Google Cloud, and paid Gemini or Gemini Enterprise Agent Platform… Read more →
Pros
- Open source (Apache-2.0) and auditable
- Large context window for big repositories
- Still supported for enterprise licences and paid API keys
Cons
- Retired for free, AI Pro/Ultra and individual accounts on 18 June 2026
- Google's development focus has moved to the Antigravity CLI
Pricing: Open source · terminal google open source
33
Free local dashboard that tracks AI coding token spend across 40+ tools
★ 11k · +66 this week · MIT · updated 2026-10-02
6.5/10
What it does Codeburn is a local-first dashboard and CLI that tracks how much your AI coding tools are actually costing you. It reads the session files that tools like Claude Code, Cursor, Codex, and GitHub Copilot already write to disk, then breaks down token usage and spend by model, project, task, and tool — with no account, no API… Read more →
Pros
- 100% local — reads existing session files with no account, API keys, or uploads; optional desktop telemetry defaults off in the EU/EEA/UK
- Covers roughly 40 tools/agents with genuinely useful features (waste detection with auto-fix, budget guards, spend-vs-commits correlation)
- Free and open source (MIT) with an active community (11k+ GitHub stars, reportedly 150k+ users)
Cons
- It's an observability/cost tool, not a coding agent — it doesn't write or edit code itself
- Cost estimates rely on a 24-hour-cached pricing feed (LiteLLM), so very new or custom models can lag briefly
Pricing: Open source · cost-tracking cli analytics mcp dashboard
34
ANUS Open source
Free terminal coding agent that auto-routes between free LLM tiers
★ 6.6k · MIT · updated 2026-10-01
6.5/10
What it does ANUS (Autonomous Networked Utility System) is a free, terminal-based coding agent that reads code, edits files, and runs commands like other CLI agents — but instead of requiring a paid API key, it automatically routes each request to whichever free LLM tier is currently available and capable. It maintains a ranked list of free models, refreshed twice… Read more →
Pros
- Costs nothing to run end to end by cycling across multiple providers' free tiers (OpenRouter, Gemini, Groq, Cerebras, Mistral) as rate limits hit
- Active project with MIT license, transparent local key storage, and a model-ranking list that refreshes twice a day
Cons
- Quality and availability depend entirely on whichever free tier is currently up, so behavior can vary day to day; some providers' free tiers use submitted data
Pricing: Open source · terminal-agent free-models cli open-source openrouter
35
Terminal coding agent for DeepSeek models with agent-to-agent delegation and review
★ 1.0k · +509 this week · MIT · updated 2026-10-02
6.5/10
What it does DSCODE is a terminal-based AI coding agent built on top of DeepSeek Harness and DSH-Code. Like Claude Code or Codex, it reads your repository, edits files, runs commands and tests, and loops until a task is done, but it is built specifically around DeepSeek models (with OpenRouter and OpenCode Go as alternative providers). Its distinguishing feature is… Read more →
Pros
- Agent-to-agent delegation with isolated worktrees and an independent code-review pass
- Supports skills, MCP, AGENTS.md, plugins and hooks
- Actively maintained with frequent releases
Cons
- macOS-only; no Windows/Linux support
- Requires a paid DeepSeek/OpenRouter API key to actually run tasks
Pricing: Open source · deepseek terminal-agent macos mcp multi-agent
36
Multi-engine desktop GUI wrapping Claude Code, Codex, OpenCode and other CLI coding agents
★ 4.4k · +45 this week · updated 2026-09-30
6.5/10
What it does Desktop CC GUI is a Tauri-based desktop application that puts a single graphical shell around multiple terminal coding-agent CLIs, so you can run Claude Code, Codex and other engines side by side without juggling separate terminal windows for each one. What is inside - Multi-engine support: Claude Code, Codex CLI, Kimi CLI, Grok CLI, Pi CLI, OMP… Read more →
Pros
- One GUI for many CLI coding agents, with file tree, terminal and Git panel built in
- Fast-moving, genuinely active project (6,000+ commits, 4.4k stars)
- Free and open source with no account or paid tier required to run the desktop client
Cons
- Very high open-issue count (299) relative to project size suggests rough edges to expect
- Breadth across ten+ engines may mean shallower polish per engine than a single-purpose wrapper
Pricing: Open source · desktop-gui multi-engine claude-code codex tauri
37
Terminal AI coding agent with persistent memory and multi-agent modes
★ 14k · +89 this week · MIT · updated 2026-09-30
6.5/10
What it does MiMo Code is a terminal-native AI coding agent from Xiaomi's MiMo team. It reads and edits code, runs shell commands, manages Git, and keeps a persistent, SQLite-backed memory of your project so it can pick up context across sessions instead of starting cold every time. It ships three agent modes: Build (full read/write/execute permissions), Plan (read-only planning),… Read more →
Pros
- Persistent, SQLite-backed project memory and automatic checkpoints keep long sessions coherent across restarts
- Provider-agnostic — works with Xiaomi's own MiMo models, OpenAI, Anthropic Claude, xAI Grok, or any custom OpenAI-compatible endpoint
- Ships 20+ built-in skills (PDF/Word/PowerPoint/Excel handling, arXiv analysis, research writing) well beyond plain code editing
Cons
- macOS support is limited to iTerm2 or the VS Code terminal — Terminal.app isn't supported
- An unrestricted permission flag exists that auto-approves every agent action if misused; the docs explicitly warn against it outside disposable environments
Pricing: Open source · terminal cli multi-agent memory mcp
38
Refly Open source
Compiles natural-language workflows into versioned, exportable agent skills
★ 7.5k · +2 this week · updated 2026-07-29
6.5/10
What it does Refly lets you describe a business workflow in plain language ("Vibe Construction") and compiles it into a versioned, auditable skill instead of a one-off prompt. The pitch is that skills should be infrastructure: stored in a registry, versioned, and exportable to multiple runtimes rather than re-written per agent. What is inside - A skill registry for centralized,… Read more →
Pros
- Compiles natural-language workflows into versioned, auditable skills instead of throwaway prompts
- Exports to APIs, webhooks, MCP servers or native tools, with 3,000+ documented connectors
- Intervenable runtime lets a human pause and audit execution mid-run
Cons
- No published hosted pricing; effectively self-hosted, so you carry infra and LLM API costs
- Broad integration claims (Claude Code, Cursor, Lovable, Slack, Lark) are documentation-level and not independently verified here
Pricing: Open source · agent-skills workflow mcp automation open-source
39
How long until a local LLM rig pays for itself?
6.5/10
What it does Sunk Cost is a free, single-page web calculator that answers a specific question: if you buy hardware to run large language models locally, how long until that purchase pays for itself compared to just paying for a cloud API? It was built after its creator kept hearing "just buy a Mac and run models locally, it pays… Read more →
Pros
- Free, no signup, no cookies or IP logging
- Grounds the "buy hardware and run locally" debate in real numbers: hardware price, power draw and token throughput
- Includes a model capability leaderboard and hardware pricing database for context
Cons
- A single-purpose calculator, not an actual coding agent or MCP tool
- Estimates depend on self-reported hardware/pricing data and usage assumptions
Pricing: Free · local-llm cost-calculator hardware break-even self-hosted
40
Whip Open source
Single-binary Go terminal coding agent tuned for open-source and OpenAI-compatible models
★ 1.1k · +16 this week · Apache-2.0 · updated 2026-09-30
6.5/10
What it does Whip is a terminal coding agent written in Go: a bubbletea TUI wrapped around an LLM tool-use loop (bash, read, write, edit, and subagent tools) that keeps working on a goal until it's done. It ships as a single static binary with no runtime to install, which makes it easy to drop onto a fresh machine, a… Read more →
Pros
- Single static binary, no runtime dependencies to install
- Provider-agnostic: works with OpenRouter, any OpenAI-compatible endpoint, or a Codex subscription
- Built-in MCP support and background subagents for concurrent work
Cons
- Young, niche project with a small contributor base compared to established terminal agents
- Documentation is still thin outside the README/docs folder
Pricing: Open source · terminal go open-source-models mcp cli
41
jcode Open source
RAM-efficient Rust coding agent harness with swarm and semantic-memory features
★ 20k · +133 this week · MIT · updated 2026-10-02
6.5/10
What it does jcode is a terminal AI coding agent harness built in Rust, whose headline claim is very low memory usage (the README reports about 27.8 MB RAM for a single session with embeddings off). It supports multi-session, multi-agent collaborative workflows against the same repository. What is inside - A semantic memory system using vector embeddings for automatic context… Read more →
Pros
- Concrete, measured RAM-efficiency claims rather than vague performance marketing
- Swarm mode plus semantic memory is a distinctive feature combination
- Broad multi-provider model support (Claude, OpenAI, Copilot, Gemini, Azure)
Cons
- Self-modifying/self-rebuilding agent mode is a real safety consideration and needs sandboxing
- High star count relative to what looks like a small/solo maintainer team warrants normal due diligence
Pricing: Open source · rust terminal-agent multi-agent embeddings self-modifying
42
Claude Code-style terminal UI plugin for DeepSeek Harness
★ 3.9k · +757 this week · MIT · updated 2026-10-02
6.3/10
What it does dsh-TUI is a terminal UI plugin for DeepSeek Harness (DeepSeek's terminal coding-agent CLI). It layers a Claude Code-style interface on top of the harness: a pixel-art whale header, live work-status and TPS display, streaming "thinking" output, double-Esc time rewind, a context-usage bar, inline image previews, and in-terminal Mermaid diagram rendering. What is inside Session management (/resume, /home,… Read more →
Pros
- Very actively maintained with CI, an English/Chinese README, and official recognition from DeepSeek Harness's own channel
- Rich terminal feature set: image/diagram previews (Kitty/Sixel, Mermaid), session management, VS Code companion extension, all as a zero-core-changes plugin
Cons
- Hard dependency on DeepSeek Harness plus a DeepSeek API key — it is not a standalone agent
- README explicitly warns the default Windows profile ships with an unrestricted-access permission mode, so review that setting before use
Pricing: Open source · terminal-ui deepseek tui-plugin vscode-integration open-source