Terminal coding agents
Command-line agents that work inside your repository: they read code, edit files, run tests and commit, driven from a terminal.
By weekly package downloads, the leaders are OpenAI Codex (25.6M), Claude Code (14.8M), Pi (5.2M). By VS Code extension installs, the leaders are Claude Code (27.1M), OpenAI Codex (15.2M), Cline (5.5M). By GitHub stars, the leaders are OpenCode (212K), Claude Code (150K), OpenAI Codex (128K). The fastest grower over the last 30 days is Oh My Pi, with downloads up 54%. As of Oct 6, 2026.
| # | Agent | Pulse | Downloads 7d | 7d | 30d | VS Code installs | Stars | Latest release | Price | Last 90 days | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 |
OpenAI CodexOpenAI |
82 | 25.6M | up 5.7% | up 21.2% | 15.2M | 128K+153/day | 0.160.1yesterday | Free + $8/mo | ||
| 3 |
Claude CodeAnthropic |
80 | 14.8M | up 11.6% | down 36.4% | 27.1M | 150K+161/day | 2.1.290yesterday | $20/mo | ||
| 5 |
PiEarendil Inc. (created by Mario Zechner) |
75 | 5.2M | up 29.8% | up 52.1% | — | 113K+378/day | 1.0.4yesterday | Free (OSS) | ||
| 7 |
OpenCodeAnomaly (SST team) |
73 | 3.3M | up 22.0% | up 6.1% | 1.2M | 212K+200/day | 1.18.346d ago | Free + $10/mo | ||
| 24 |
Gemini CLIGoogle |
57 | 453K | up 4.7% | up 1.8% | 1.1M | 107K+24/day | 0.62.07d ago | Varies | ||
| 26 |
Oh My PiStencil Labs (Can Bölük) |
56 | 241K | up 11.4% | up 53.9% | — | 34.4K+111/day | 18.6.12d ago | Free (OSS) | ||
| 34 |
ClineCline Bot Inc. |
52 | 95K | up 10.1% | down 37.2% | 5.5M | 69.9K+63/day | 0.0.434d ago | Free + $9.99/mo | ||
| 40 |
OpenHandsOpenHands (formerly All Hands AI) |
50 | 316K | — | — | — | 90.1K+98/day | 1.25.0today | Free | ||
| 48 |
mini-SWE-agentPrinceton University & Stanford University (SWE-agent team) |
47 | 3.4M | up 58.9% | down 26.3% | — | 8,225+23/day | 2.4.62mo ago | Free (OSS) | ||
| 51 |
Kilo CodeKilo Code (Anaconda) |
46 | 32.8K | down 32.4% | — | 1.6M | 27.5K+15/day | 7.8.35d ago | Free + $19/mo | ||
| 52 |
GitHub Copilot CLIGitHub (Microsoft) |
46 | 1.7M | up 19.1% | down 36.3% | — | 11.2K+5/day | 1.0.92yesterday | Free + $10/mo | ||
| 53 |
Grok BuildxAI (SpaceXAI) |
44 | 74.2K | up 14.8% | up 22.7% | — | 27.2K+18/day | — | — | ||
| 58 |
GooseBlock (now Agentic AI Foundation) |
43 | — | — | — | 7,358 | 55K+38/day | 1.53.04d ago | Free (OSS) | ||
| 65 |
Qwen CodeAlibaba Cloud (Qwen) |
41 | 105K | up 30.8% | up 5.5% | 373K | 28.3K+20/day | 0.25.0yesterday | Bring your key | ||
| 84 |
AiderAider-AI (Paul Gauthier) |
34 | 61.3K | down 1.5% | down 57.8% | — | 49.4K+23/day | 0.86.014mo ago | Free (OSS) | ||
| 86 |
Kimi CodeMoonshot AI |
34 | 43.7K | up 21.9% | down 8.3% | 598K | 7,775+8/day | 2.1.112d ago | — | ||
| 88 |
Mistral VibeMistral AI |
34 | 488K | down 30.9% | down 60.1% | 40.8K | 5,050+5/day | 2.25.813d ago | Free + $14.99/mo | ||
| 90 |
CrushCharm |
33 | 7,773 | down 10.6% | — | — | 28.5K+25/day | 0.97.17d ago | Free | ||
| 92 |
Antigravity CLIGoogle |
33 | — | — | — | — | — | — | Free | ||
| 96 |
CodebuffCodebuffAI |
31 | 1,480 | up 50.3% | — | — | 13.2K+47/day | — | $100/mo | ||
| 97 |
AmpAmp (spun out of Sourcegraph) |
31 | 41.2K | up 7.9% | up 32.7% | — | — | — | Free + $20/mo | ||
| 107 |
Open InterpreterOpen Interpreter (openinterpreter.com) |
26 | — | — | — | — | 68.5K+12/day | 0.0.556d ago | Free (OSS) | ||
| 112 |
Kiro CLIAmazon Web Services |
22 | — | — | — | — | — | — | Free + $20/mo | ||
| 113 |
Cursor CLICursor |
21 | — | — | — | — | — | — | Free + $20/mo | ||
| 131 |
DroidFactory |
16 | 9,261 | down 11.5% | down 23.3% | 39.5K | 44+1/day | — | $20/mo | ||
| 155 |
Rovo Dev CLIAtlassian |
— | — | — | — | — | — | — | $20/mo | — | |
| No agents match that filter. | |||||||||||
Ranks are overall positions by Pulse Score. Click a column to sort by what matters to you. Methodology.
Why terminal agents took off
A command-line agent needs nothing but a shell, so it works with any editor, over SSH, in containers and in CI. It can run the same commands you do (tests, linters, build tools, git), and you can script it: pipe a failing test log in, ask for a fix, review the diff.
What separates them
- Model access. Some agents are tied to one vendor's models and bundled with that vendor's subscription. Open-source agents usually let you bring your own API key or point at a local model.
- Permissions. How the agent asks before running commands or editing files, and whether it can run in a sandbox.
- Extensibility. Support for the Model Context Protocol (MCP), custom commands, hooks and sub-agents.
- Cost model. A flat subscription with usage limits, or pay-per-token through an API key. Heavy users should compare both.
Reading the numbers
These tools are distributed mostly through npm, PyPI and Homebrew, so downloads are a direct signal here. Keep one caveat in mind: agents that auto-update or ship daily releases are downloaded more often per user than those that ship monthly.
Head-to-head comparisons
Related guides
Best AI Agents in 2026: Live Rankings by Category
The leading AI agents for coding, automation, research, browsing, app building, customer support, voice and sales, ranked by public usage data that refreshes daily, with prices and free options.
Best AI Coding Agents in 2026: Ranked by Usage, Benchmarks and Price
The AI coding agents developers use most, with live download and install numbers, benchmark results, entry prices and a way to choose between terminal, IDE and cloud agents.
Claude Code vs Codex vs Cursor: Usage, Price and Benchmarks Compared
The three most discussed AI coding agents side by side on live data: downloads, installs, GitHub activity, release cadence, benchmark results and pricing.
How to Start Using a Terminal Coding Agent: Setup, First Tasks and Guardrails
A practical manual for your first week with a command-line coding agent: installing one, running a first session, writing project instructions, setting permissions and keeping costs under control.
Questions about terminal coding agents
What is the most downloaded terminal coding agent?
OpenAI Codex, with 25.6M package downloads in the last 7 days, ahead of Claude Code (14.8M) and Pi (5.2M).
Can I use a terminal coding agent with my own API key or a local model?
Most open-source terminal agents support this, and several commercial ones accept an API key as an alternative to a subscription. Check the "Bring your own API key" row on each agent's page.
Is a terminal agent better than an AI IDE?
Neither is better in general. Terminal agents are easier to automate and work with any editor; AI IDEs make it easier to see and review what the agent changed. Many developers use both.