Agents that edit real codebases. Ranked by our fixed rubric. Untested tools are listed by release date with a provisional score.
These aren't ranked. Their verdicts come from public sources only.
Claude Code 2.1.300Verdict: An incremental release aimed at people customizing Claude Code with mods and monitoring it with OpenTelemetry. Nothing changes for everyday use.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Paid (needs a Claude plan or API key)
Kilo Code 7.8.10Verdict: A small quality-of-life release. Useful if your agent kept missing generated files you actually wanted it to read.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (you pay for model usage)
Vani Code Agent for VS CodeVerdict: A careful, approval-first coding agent and an easy pick for VS Code users who want to try another model. It's early-stage: desktop apps and a CLI haven't shipped yet.
Test notes: Not tested: it needs an account, a paid plan or a heavy local setup. The verdict is based on the linked public sources only.
Pricing: Free extension; usage billed via API key
Cline 4.1.24Verdict: A maintenance release: mostly model-catalog and default changes. Check your model selection after updating, since defaults changed for many providers.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (you pay for model usage)
Gemini 4 Argon (incl. GitHub Copilot)Verdict: It's the new model to try for long refactors and migrations if you already pay for Copilot or AI Ultra. Access on lower Copilot tiers is limited.
Test notes: Not tested: it needs an account, a paid plan or a heavy local setup. The verdict is based on the linked public sources only.
Pricing: Paid (Gemini API paid tier, AI Ultra, Copilot Pro+ and up)
Gemini CLI 0.64.0Verdict: A maintenance release, mostly relevant if you run Gemini CLI as an A2A server or through ACP-based editors.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (free tier with a Google account; paid API optional)
Muse Wonder / Muse Wonder FullVerdict: A strong new contender for heavy, multi-file agent work. The best version sits behind the Power and Maximum Usage plans.
Test notes: Not tested: it needs an account, a paid plan or a heavy local setup. The verdict is based on the linked public sources only.
Pricing: API $4/$20 per 1M tokens; Full via subscription
OpenHands 1.27.1Verdict: A patch release. Update if you run OpenHands in Docker.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (you pay for model usage)
OpenCode 1.19.0Verdict: Mainly a bridge release to OpenCode 2, plus a PTY security check. Run /update when you're ready, since auto-updates are now off.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (you pay for model usage)
goose 1.54.0Verdict: Shareable HTML session exports are handy for reviewing or demoing what an agent did. A solid pick if you want a model-agnostic local agent.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (you pay for model usage)
OpenAI Codex CLI 0.162Verdict: Worktree support is the headline: it makes running parallel agent tasks on one repo cleaner. Worth updating if you run Codex on several tasks at once.
Test notes: Not tested: the verdict is based on the linked public release notes only.
Pricing: Open source (needs a ChatGPT plan or API key)
The week's new agent tools with a verdict on each, every week. Free. We'll email you a link to confirm, and you can unsubscribe any time. See our Privacy Policy.