GitHub's AI pair-programmer — inline completions, chat, and the new Agent mode that ships PRs.
CodeTool useMemory
Demo · hover to play
24 Agenten · Capability-Hub
KI-Agenten mit Codeausführung2026
Agenten, die Code in einer Sandbox-Umgebung schreiben, ausführen und iterieren. Erforderlich für Datenarbeit, Scripting und jede Aufgabe, die mit „und verifizieren, dass es wirklich funktioniert" endet.
Möchten Sie die technische Definition? Zum Glossar-Eintrag „Codeausführung" →
Die 24 Agenten mit Codeausführung
Codeium's AI editor — Cascade agent flows alongside in-line completion and chat.
CodeTool useMemoryBrowserDemo · hover to playBackground agent that drives the Cursor editor across multi-file changes.
CodeTool useMemoryDemo · hover to playVercel's generative UI agent — design and ship React components from natural language.
CodeTool useVisionDemo · hover to playAutonomous AI software engineer that ships PRs end-to-end.
CodeTool useBrowserMemoryDemo · hover to playMicrosoft's open-source multi-agent framework — orchestrate conversations between specialised agents.
Multi-agentTool useCodeMemoryAnthropic's terminal agent — composable, scriptable, and built around Claude's tool-use loop.
CodeTool useMemoryDemo · hover to playVibe-coding builder for non-engineers — prompt a full-stack app and ship it to a live URL in minutes.
CodeTool useVisionDemo · hover to playStackBlitz's in-browser AI builder — generates and deploys real Node.js apps from a single prompt.
CodeTool useVisionDemo · hover to playGeneral-purpose agent that turns a single prompt into a finished deliverable.
BrowserTool useCodeMemoryGet AGENTS20codeAGENTS20
Demo · hover to playBuild and ship full applications from a single prompt — runs in the Replit cloud.
CodeTool useBrowserMemoryDemo · hover to playAI dev tool that builds entire production apps from spec — formerly GPT Pilot, open-source roots.
CodeTool useMemoryDemo · hover to playAWS's enterprise AI agent — coding assistant, business chat, and customizable Q apps grounded in your AWS data.
CodeTool useRAGMemoryDemo · hover to playAutonomous coding agent grounded in your whole codebase — multi-file edits with deep context.
CodeTool useMemoryRAG
Demo · hover to playPrivacy-first AI coding assistant — runs on your own infrastructure or in a private cloud.
CodeTool useMemoryDemo · hover to playOSS terminal coding assistant — pair-programs, edits in place, commits as you go.
CodeTool useOpen-source autonomous coding agent that lives in your IDE.
CodeTool useBrowserDemo · hover to playCode intelligence agent for enterprise — answers questions about your code, navigates large monorepos.
CodeTool useRAGMemoryDemo · hover to playByteDance's AI coding IDE with autonomous Builder mode — multi-file scaffolding from a single prompt.
CodeTool useMemoryOpen-source coding agent for VS Code and JetBrains — bring your own model, customize every prompt.
CodeTool useMemoryDemo · hover to playOpenAI’s open-source terminal agent for refactors, audits and migrations.
CodeTool useDemo · hover to playGitHub's natural-language micro-app builder — describe a personal app, get a deployed working version.
CodeTool useMemory
Demo · hover to playAutonomous AI engineer trained on long contexts — handles hour-long coding sessions without losing track.
CodeTool useMemory
Demo · hover to playGitHub-native agent that turns issues into reviewed pull requests.
CodeTool useMemoryDemo · hover to play
Häufig gefragt
Was bedeutet Codeausführung bei KI-Agenten?+
Die Fähigkeit eines Agenten, Code in einer kontrollierten Umgebung zu schreiben, auszuführen und basierend auf dem Ergebnis weiter zu reasonieren.
Welche KI-Agenten unterstützen Codeausführung?+
24 Agenten in unserem Index bieten Codeausführung. Die Liste oben ist nach Community-Interesse sortiert; GitHub Copilot, Windsurf, Cursor Agent sind 2026 die am meisten recherchierten.
Wie bewerte ich Codeausführung bei einem KI-Agenten?+
Achten Sie auf: (1) Zuverlässigkeit bei Edge Cases, nicht nur in Demo-Videos; (2) wie der Agent sich erholt, wenn Codeausführung mitten in der Aufgabe ausfällt; (3) ob Codeausführung der Standardmodus oder ein Opt-in-Feature ist. Produktionsreife Agenten veröffentlichen Codeausführung-Benchmarks; Demos und Screenshots reichen nicht.
Weitere Fähigkeiten entdecken