10+ Agents
Claude Code · Cursor · Copilot · Codex · Gemini CLI · Devin · Windsurf · Amp · AdaL
codingagent.md
The coding agent ecosystem, documented.
5+ Config Formats
AGENTS.md · CLAUDE.md · .cursorrules · copilot-instructions.md · SKILL.md
2 Protocols
MCP (Model Context Protocol) · ACP (Agent Client Protocol)
60K+ Repos
Already shipping AGENTS.md files across the open-source ecosystem
Navigate the ecosystem
Section titled “Navigate the ecosystem” Coding Agents Profiles and comparisons of every major AI coding agent
Config Formats How to write AGENTS.md, CLAUDE.md, .cursorrules, and more
Protocols MCP vs ACP — what they do and how they compare
Guides Getting started, choosing an agent, multi-agent setups
Models powering the agents
Section titled “Models powering the agents”Which LLMs perform best for coding? We track benchmarks weekly. Full rankings →
| Model | Provider | SWE-bench Verified | Context | Best For |
|---|---|---|---|---|
| Claude Opus 4.6 | Anthropic | 80.8% | 200K | Complex refactors, agentic coding |
| GPT-5.2-Codex | OpenAI | 80.0% | 128K | Terminal workflows, long-horizon tasks |
| Claude Opus 4.5 | Anthropic | 80.9% | 200K | Deep reasoning, legacy code |
| Gemini 3 Flash | 78.0% | 1M | Large codebases, best value | |
| Claude Sonnet 4.5 | Anthropic | 77.2% | 200K | Day-to-day development |
| Gemini 3 Pro | ~75% | 1M | Multimodal, massive repos | |
| DeepSeek V3 | DeepSeek | 42.0% | 128K | Budget-friendly |
| Qwen 2.5 Coder 32B | Alibaba | — | 128K | Local / privacy-first |
Stay updated
Section titled “Stay updated”Get weekly updates on the coding agent ecosystem — new agents, format changes, benchmarks.