State of Coding Agents
Industry report
Last updated: July 2026

State of Coding Agents
Mid-2026

A data-driven overview of the 28 AI coding agents shaping how software gets built: who makes them, where they run, how they interoperate, and where the ecosystem is heading.
28Agents indexedAcross the landscape
27VendorsxAI owns two agents
14 / 14Open / ClosedAn even split
100%Support AGENTS.mdUniversal standard
96%Support MCPNear-universal
14New in 2025The breakout year
Timeline

The 2025 explosion

2025 was the breakout year. Half of all agents indexed (14 of 28) launched in a single year, driven by terminal agents and big-tech entrants.
2025 alone produced 14 new agents, as many as 2021 through 2024 combined. The pace has not slowed: 2026 is already on the board with four more.
2021 (1)
GitHub Copilot launches as the first major AI coding assistant.Released this year
  • GitHub Copilot
2023 (4)
Cursor and Zed arrive as IDE-first approaches begin.Released this year
  • Factory
  • Cursor
  • Zed
  • Aider
2024 (5)
Devin launches as the first "AI software engineer," an autonomous cloud agent.Released this year
  • Devin
  • OpenHands
  • Augment Code
  • Warp
  • Cline
2025 (14)
The breakout year. Claude Code accelerates the terminal-agent wave; OpenAI ships Codex; Google enters with Antigravity. Cognition AI acquires Windsurf.Released this year
  • Junie
  • Trae
  • Goose
  • Claude Code
  • VS Code (agent mode)
  • Kilo Code
  • Codex
  • Amp
  • opencode
  • Kiro
  • Qwen Code
  • Antigravity
  • Mistral Vibe
  • Kimi Code CLI
2026 (YTD) (4)
Zhipu AI, xAI, and Xiaomi ship agents as even more big-tech entrants arrive. xAI acquires Cursor.Released this year
  • Command Code
  • Grok Build
  • MiMo Code
  • ZCode
Where They Live

The terminal is the universal surface

Agents can span multiple surfaces. 25 of 28 ship a terminal/CLI interface, but the trend is multi-surface: 19 of 28 (68%) run on two or more, and 14 span three or more.
Terminal
25 · 89%
IDE Extension
12 · 43%
Web
11 · 39%
Desktop
10 · 36%
IDE (full)
7 · 25%
Multi-surface is the direction of travel. The terminal is the lowest common denominator, but the leading platforms increasingly span editor, web, and desktop too.
Terminal-only purists (6)CLI-first workflows, nothing else.
  • Aider
  • Amp
  • Command Code
  • Grok Build
  • Mistral Vibe
  • Qwen Code
Most surfaces, 4 categories (4)Meet developers wherever they are.
  • Augment Code
  • Claude Code
  • Devin
  • Factory
IDE-only (2)Editors that add agents natively.
  • VS Code
  • Zed
Standards Stack

A shared interoperability stack is emerging

Seven standards are converging across the ecosystem. AGENTS.md is universal, MCP and Skills are near-universal, and even newer standards like hooks are catching up fast.
Rules (AGENTS.md)Repo-level instruction files agents read for project context and conventions
28/28 · 100%
SkillsReusable, packaged skill modules that extend agents with task-specific capabilities
27/28 · 96%
MCPModel Context Protocol, an open standard for connecting agents to external tools and data
27/28 · 96%
Custom Models / BYOKSupport for plugging in custom, self-hosted, or bring-your-own-key model endpoints
24/28 · 86%
SubagentsSupport for spawning or orchestrating sub-agents to delegate tasks within workflows
24/28 · 86%
PluginsPlugin/extension systems for third-party integrations and custom tools
22/28 · 79%
HooksLifecycle hooks that trigger custom behavior in response to agent events
21/28 · 75%
AGENTS.md is the de facto standard. 27 of 28 agents read AGENTS.md for project context. 17 agents support all seven standards; the laggards are Aider (2/7), Warp (3/7), Zed (3/7), and VS Code (4/7).
Interoperability

Standard convergence, implementation divergence

Agents agree on AGENTS.md, but beyond that each reads its own sprawl of proprietary rule files. .agents/skills/ is winning as the shared path for skills, yet 27 agents collectively maintain 55 distinct skills directories.
Rules file
AGENTS.md27 of 28
Proprietary rules
.cursor/rules, .windsurfrules, …28 of 28 (each has its own)
Skills directory
.agents/skills/19 of 28 (55 distinct paths)
Claude rules compat
CLAUDE.md10 of 28 (cross-compatible)
MCP config
.mcp.json3 (emerging)
Everyone reads AGENTS.md — then diverges. 27 of 28 agents read AGENTS.md for project context, but each also reads its own proprietary instruction files. .agents/skills/ is the closest thing to a shared skills path, yet the 27 agents that support skills collectively use 55 distinct directory paths. The standard is settled; the implementation is not.
Open vs Closed

An exact 50/50 split

Fourteen agents are open source and fourteen are closed, a remarkably even balance for a fast-moving market.
Open vs closed source
28agents
Open source 14 (50%)
Closed source 14 (50%)
OSS license breakdown
14OSS
Apache-2.0 7 (50%)
MIT 5 (36%)
AGPL-3.0 1 (7%)
GPL-3.0 1 (7%)
GitHub stars: open-source repos (July 2026)
VS Code
187k
opencode
180k
Claude Code
135k
Codex
94k
Zed
86k
OpenHands
79k
Cline
64k
Warp
63k
Goose
50k
Aider
47k
Qwen Code
26k
Kilo Code
25k
Mistral Vibe
5k
Open at the edges. Several closed-source agents still publish their CLI as open source (for example, Claude Code's repo on GitHub) while the cloud platform stays proprietary. The boundary between open and closed is blurrier than the 50/50 split suggests.
Big Tech vs Startups

Big tech edges ahead of startups, 15 to 13

Every major AI lab now ships a coding agent (OpenAI, Anthropic, Google, xAI, Mistral, Amazon) alongside Chinese labs Alibaba, Xiaomi, ByteDance, Moonshot AI, and Zhipu AI. xAI owns two: Cursor and Grok Build.
Big tech & major AI labs (15)
AntigravityGoogle
Claude CodeAnthropic
CodexOpenAI
CursorxAI
GitHub CopilotGitHub
Grok BuildxAI
JunieJetBrains
Kimi Code CLIMoonshot AI
KiroAmazon
MiMo CodeXiaomi
Mistral VibeMistral AI
Qwen CodeAlibaba
TraeByteDance
VS CodeMicrosoft
ZCodeZhipu AI
Independent & startup (13)
AiderAider-AI
AmpAmp
Augment CodeAugment Code
ClineCline
Command CodeCommandCodeAI
DevinCognition AI
FactoryFactory
GooseAgentic AI Foundation (Linux Foundation)
Kilo CodeKilo Code
opencodeAnomaly
OpenHandsAll Hands AI
WarpWarp
ZedZed Industries
Coding agents are now essential infrastructure. The fact that every frontier lab has shipped one, and that five Chinese labs are all in, signals the category has moved from experiment to strategic priority.
State of Coding Agents
A factual, data-driven overview of the AI coding agent landscape. No rankings, no sponsorships, just the numbers. Figures are compiled from public vendor documentation and GitHub as of July 2026 and are approximate.
Last updated: July 2026