~ / agents

AI Coding Agents

A ranked directory of coding agents, from terminal-first tools to IDE assistants and browser-based builders.

Claude Code

deep terminal workflows and complex multi-step refactors

score 86 / CLI/Desktop

ChatGPT Codex

ChatGPT Codex workflows, terminal automation, and CI-style editing

score 84 / CLI/Desktop

Cursor

daily IDE coding with autocomplete and composer workflows

score 71 / IDE

Devin

long-running autonomous cloud tasks

score 61 / Cloud

ClineOS

MCP-heavy VS Code workflows and bring-your-own-model setups

score 61 / Extension

AiderOS

git-native terminal editing and experienced open-source users

score 60 / CLI

All tracked agents

Sorted by architecture score.

#AgentScoreTypePriceBest fit
1Claude Code86CLI/Desktop$20/$200deep terminal workflows and complex multi-step refactors
2ChatGPT Codex84CLI/Desktop$20/$200ChatGPT Codex workflows, terminal automation, and CI-style editing
3Cursor71IDE$20/mo+daily IDE coding with autocomplete and composer workflows
4Devin61Cloud$500/molong-running autonomous cloud tasks
5ClineOS61ExtensionFree agentMCP-heavy VS Code workflows and bring-your-own-model setups
6AiderOS60CLIFree agentgit-native terminal editing and experienced open-source users
7Windsurf59IDE$15/mo+AI-native IDE workflows with Cascade-style multi-file edits
8ContinueOS59ExtensionFree agentcustom context providers across VS Code and JetBrains
9OpenCodeOS53CLIFree agentprovider-flexible terminal coding with open-source control
10Copilot52IDE$10/mo+GitHub-centered teams and IDE-native suggestions
11Oh-My-CodexOS47CLIFree agentopen-source Codex enhancements
12Qwen CodeOS45CLIFree agentQwen-first open-source CLI workflows
13GooseOS43CLIFree agentMCP-compatible local automation
14Gemini CLI40CLIFreelegacy Google CLI workflows
15Hermes AgentOS39CLIFree agentskills-based terminal experimentation
16Amazon Q Developer38IDESubscriptionAWS-heavy enterprise teams
17Antigravity CLI35CLITBDGoogle terminal-agent exploration
18Pi AgentOS35CLIFree agentminimal-token terminal use
19Replit Agent34BrowserSubscriptionbrowser-based app building
20Tabnine28IDE$12/mo+enterprise code completion and private deployments
21Bolt.new25BrowserSubscriptionbrowser-based full-stack prototypes

Latest bots

Always-on agents shipped since late September 2026. Not yet scored — open a card for the full feature, setup, use-case and download profile.

Meta Muse

Meta’s first personal AI agent app (8 Sep), now live in the US, Japan and Singapore; the Muse Wonder models answer behind it.

not scored / Mobile+Web

Grok Bot

xAI beta: each bot gets its own cloud computer — browser, files, terminal — and runs multi-step jobs 24/7 under SuperGrok Heavy.

not scored / Cloud

DeepSeek Super-agentOS

MIT open-source always-on coding agent from the Harness team; finds defects unprompted in group chats and runs fixes in parallel.

not scored / CLI

OpenAI Dots

Always-on agents in ChatGPT on GPT-6 Astra; each Dot gets its own cloud computer, browser and 4,000+ app plugins.

not scored / Cloud

Gemini Agent

Google Cloud’s universal work agent (8 Oct): one prompt box with org context, planning and running work across Workspace and M365.

not scored / Cloud

Factory Droid 1.0

Model-independent harness; script-mode tool calling turns one model request into hundreds of sandboxed tool calls.

not scored / CLI

China big-tech entries

Not yet scored — feature profiles only.

Trae

ByteDance VS Code-fork IDE with Builder + SOLO; open-source Trae-Agent CLI.

not scored / IDE+CLI

CodeBuddy

Tencent coding arm (Buddy AI); first Chinese assistant with MCP; Craft + Plan.

not scored / IDE+CLI

WorkBuddy

Tencent desktop workplace agent; local files, PPT, Excel, web automation, Skills.

not scored / Desktop

Comate

Baidu ERNIE assistant (2023); Zulu agent runs requirement-to-code; CAICT 4+.

not scored / Plugin+IDE

Qoder

Alibaba first-party IDE; Repo Wiki codebase map + autonomous Quest tasks.

not scored / IDE+CLI

Zcode

Z.ai desktop ADE, official GLM-5.2/5.3 harness; Goal mode, multi-agent.

not scored / Desktop

DeepSeek HarnessOS

DeepSeek MIT open-source harness; Cordis plugin kernel; provider-agnostic.

not scored / CLI

These seven carry feature parity with Codex/Claude Code but no independent DeepSWE or Terminal-Bench score, so they sit outside the ranked directory above. See the full China comparison for the use-case matrix and official links.