Here is the complete compiled list and comparison of the major players.

1. AI-Native IDEs (The "Cursor" Category)

These are forks of VS Code (or standalone Rust/Native apps) where the AI is the core of the experience, not just a plugin.

Tool Primary Philosophy Key Feature Pricing (Est. 2026)

Cursor The Gold Standard Composer & Agent Mode: Massive multi-file edits with "predictive" diffing. $20/mo (Pro)

Windsurf Context-First Cascade: An agent that "watches" your workflow and proactively fetches context. $15/mo (Pro)

Trae Radical Access Adaptive Mode: Auto-configures environments and installs missing libs for you. Free (ByteDance backed)

PearAI Open-Source Clone Open Compliance: A fork of Cursor that lets you use your own API keys (BYOK). Free / BYOK

Zed Performance & Speed Native Rust Engine: Extremely fast; hosts external agents via the ACP protocol. Free / Pro tiers

2. Standalone Agents (The "Antigravity" Category)

These tools often work alongside your editor or in the terminal. They are better at autonomous "missions" (e.g., "Build me a login page with Auth0 and tests").

Tool      Primary Philosophy  Best For            Standout Detail

Google Antigravity     Manager View               Multi-tasking Spawns parallel agents in different "workspaces" to work on separate Jira tickets.

Claude Code Pure Reasoning           Deep Bug Fixing          CLI-based; highest scores on coding benchmarks for logic and architectural fixes.

Aider   Git-Native        Refactoring    Works in the terminal; automatically commits changes with excellent summaries.

Cline (ex-Claude Dev)             Open Control Full Automation          VS Code extension that can run browser tests and terminal commands autonomously.

Roo Code        Reliability        Complex Legacy Code            A fork of Cline focused on avoiding "agent thrashing" (endless loops of errors).

ToolPrimary PhilosophyBest ForStandout DetailGoogle AntigravityManager ViewMulti-taskingSpawns parallel agents in different "workspaces" to work on separate Jira tickets.Claude

CodePure ReasoningDeep Bug FixingCLI-based; highest scores on coding benchmarks for logic and architectural fixes.AiderGit-NativeRefactoringWorks in the terminal; automatically commits changes with excellent summaries.Cline (ex-Claude Dev)Open ControlFull AutomationVS Code extension that can run browser tests and terminal commands autonomously.Roo CodeReliabilityComplex Legacy CodeA fork of Cline focused on avoiding "agent thrashing" (endless loops of errors).

3. Integrated Platform Tools

The "Pragmatic" choice for teams already locked into a specific ecosystem (GitHub, AWS, Google Cloud).

• GitHub Copilot (Agent Mode): Now supports "Copilot Workspaces." Best for enterprise teams who want AI-integrated PR reviews and issue tracking. ($10–$39/mo)
• 
  
• Amazon Q Developer: Dominates in "Instructional Persistence." It rarely forgets rules during long tasks and is the best for AWS infrastructure (Terraform/CDK). (Free tier available)
• Gemini Code Assist: Features a 1M+ Token Context Window. This is the only tool that can effectively "read" a million-line monolithic codebase in one go.

Comparison: Which should you choose?

Go with Cursor if...

You want the most polished, "magical" feeling experience where the AI feels like it's reading your mind as you type. It is currently the most popular for a reason.

Go with Antigravity or Claude Code if...

You are more of a "Product Manager" than a "Coder." These tools are better at taking a high-level requirement and returning 10 minutes later with a finished feature.

Go with Windsurf or Trae if...

You are cost-conscious. Windsurf offers a similar experience to Cursor for $5 less, and Trae is currently free while it builds its user base.

Go with Cline / Roo Code if...

You are a power user who wants to pick the exact model (GPT-5, Claude 4, or a local Llama model) to save money or keep code strictly private.