11. AI Coding Agents Matrix
URL: coding-agents-matrix.dev | GitHub: PackmindHub/coding-agents-matrix | License: Apache-2.0
Maintainers: Packmind (Cédric Teyton, Arthur Magne)
What Is It?
An interactive comparison matrix of 23 AI coding agents across 11 technical criteria:
| Category | Criteria |
|---|---|
| Identity | Open Source status, GitHub stars, first release date |
| Packaging | CLI, Dedicated IDE, IDE Extension, BYO LLM, MCP Support |
| Features | Custom Rules, AGENTS.md, Skills, Commands, Subagents, Plan Mode |
Agents compared: Aider, Claude Code, Cursor, GitHub Copilot, Continue, Goose, Windsurf, and 16 others.
Why It's Useful
Discovery tool: When you're choosing which coding agent to adopt, the Matrix helps you filter by specific technical requirements:
- "Show me open source CLI agents with MCP support"
- "Which agents support the AGENTS.md standard?"
- "Compare Claude Code vs Cursor features side-by-side"
Objective data: No marketing fluff, just feature presence/absence (Yes/No/Partial). Community-driven updates via GitHub issue templates.
Complementarity with This Guide
| Matrix (Discovery) | This Guide (Mastery) |
|---|---|
| "Which agents exist?" | "How to use Claude Code effectively?" |
| Feature comparison (11 criteria) | Workflows, architecture, TDD/SDD methodologies |
| 23 agents × shallow | 1 agent × deep (19K lines) |
| Technical specs | Practical templates (120), quiz (264 questions) |
Use case: Use Matrix to discover and compare → Choose Claude Code → Use this guide to master it.
Interactive Features
- Sortable columns: Click any criterion to sort ascending/descending
- Multi-filter: Combine filters with AND logic (e.g., "Open Source + MCP Support + Plan Mode")
- Search: Find agents by name, type, or description
- Community-driven: Propose new agents/criteria via GitHub issues
Limitations
- Snapshot, not live: Agents evolve, criteria change. Verify data freshness (last updated: Jan 19, 2026).
- Presence/absence only: Doesn't explain how features work or quality differences.
- Example: "Claude Code has Plan Mode" (Yes) vs "How Plan Mode works in practice" (not covered)
- No workflows: Doesn't teach you how to use the agents effectively (that's what this guide does).
- No performance metrics: Doesn't benchmark speed, accuracy, or cost.
Related Resources
- Packmind: Context engineering & governance for AI coding agents
- Packmind OSS: Framework for versioning AI coding context
- Context-Evaluator: Open-source scanner for CLAUDE.md / AGENTS.md / copilot-instructions.md: 17 evaluators (13 error detectors + 4 suggestion generators), supports Claude Code, Cursor, Copilot, OpenCode, Codex. CLI + local web UI. Apache-2.0. The separately hosted web app documented upstream was unreachable during the 2026-09-02 check.
- Awesome Claude Code: Curated tool library
Positioning: Matrix complements this guide by helping you choose the right agent. Once you choose Claude Code, this guide teaches you how to master it.
See also: Agent Tools: Beyond Claude Code for full writeups on Hermes Agent, Codex CLI, Aider, Devin, SWE-agent, CrewAI, LangGraph, and AutoGen, plus a decision framework by use case.