AI coding tools directory

Coding agents, vibe coding apps and agent tools, rated Ship / Try / Skip

Real tools with verified links. Verdicts are provisional (based on release notes, not hands-on tests yet).

Codex CLI 0.162 (SHIP, Coding agent): Oct 8 release: managed git worktrees, pinned tasks, sandbox fixes.

Cursor CLI v2026.10.08 (SHIP, Coding agent): Safer Auto-review deletes, Opus/Sonnet 5.5 on Bedrock, skill-limit warnings.

Cursor (SHIP, AI IDE): Oct 13: Origin App Marketplace with 20+ deploy/CI/review apps.

Claude Code 2.1.300 (SKIP, Coding agent): Oct 14 fix release; auto-update covers it, nothing to act on.

Meta Muse Wonder (TRY, Model): Oct 12: frontier coding/agent model, 1M context, $4/$20 per M tokens. Benchmarks vendor-reported.

JetBrains Mellum2.1 (TRY, Model): Oct 12: open 12B MoE (2.5B active) coding model, runs in Ollama/LM Studio.

Windsurf (TRY, AI IDE): Agentic IDE. Not yet tested by us.

Zed (TRY, AI IDE): Fast editor with built-in agent panel. Not yet tested.

GitHub Copilot (TRY, Coding agent): Default pick if you live in GitHub. Not yet tested.

Cline (TRY, Coding agent): Open-source agent in VS Code, bring your own model. Not yet tested.

Aider (TRY, Coding agent): Terminal pair-programmer with git-native edits. Not yet tested.

OpenHands (TRY, Coding agent): Open-source autonomous dev agent. Not yet tested.

Devin (TRY, Coding agent): Hosted autonomous engineer. Not yet tested.

Bolt.new (TRY, App builder): Prompt-to-full-stack app in the browser. Not yet tested.

v0 (TRY, App builder): Prompt-to-UI/app from Vercel. Not yet tested.

Ollama (SHIP, Local models): Easiest way to run local models like Mellum2.1.

LM Studio (TRY, Local models): GUI for local models. Not yet tested.

CodeRabbit (TRY, Code review): AI PR review; now in Cursor's Origin marketplace.

Greptile (TRY, Code review): Codebase-aware PR review; in Origin marketplace.

E2B (TRY, Agent infra): Sandboxes for running agent code; supported by Cursor cloud agents.

Daytona (TRY, Agent infra): Dev sandboxes for agents; supported by Cursor cloud agents.

Modal (TRY, Agent infra): Serverless GPU/compute; supported by Cursor cloud agents.

Supabase (SHIP, Backend): Default Postgres + auth backend for vibe-coded apps.

Firecrawl (TRY, Agent tools): Web scraping to LLM-ready data. Not yet tested.

Browserbase (TRY, Agent tools): Hosted headless browsers for agents. Not yet tested.

LangGraph (TRY, Agent framework): Stateful agent orchestration. Not yet tested.

CrewAI (TRY, Agent framework): Multi-agent framework. Not yet tested.