Configs2026年7月14日·1 分钟阅读

Kilocode — Open-Source AI Coding Agent Platform

An open-source agentic engineering platform that supports multiple LLM providers and integrates with VS Code and JetBrains IDEs for AI-assisted code generation, editing, and debugging.

Agent 就绪

先审查再安装

这个资产需要先审查。复制的指令会要求 Agent dry-run、列出写入项,确认后再继续。

Needs Confirmation · 66/100策略:需确认
Agent 入口
任意 MCP/CLI Agent
类型
Skill
安装
Single
信任
信任等级:Established
入口
Kilocode
先审查命令
npx -y tokrepo@latest install 7c2d56e6-7fc2-11f1-9bc6-00163e2b0d79 --target codex

先 dry-run,确认写入项后再运行此命令。

Introduction

Kilocode is an open-source AI coding agent that runs inside VS Code and JetBrains IDEs. It connects to multiple LLM providers including Claude, GPT, Gemini, and local models, giving developers an AI pair programmer that can read, write, and refactor code across entire projects. The project focuses on being provider-agnostic and community-driven.

What Kilocode Does

  • Provides an agentic coding assistant inside VS Code and JetBrains IDEs
  • Supports Claude, GPT, Gemini, DeepSeek, and local LLMs via configurable providers
  • Reads and edits files across your project with full codebase context
  • Executes terminal commands and runs tests as part of its workflow
  • Offers chat-based interaction with inline code suggestions and multi-file edits

Architecture Overview

Kilocode is built as a TypeScript IDE extension that communicates with LLM APIs through a provider abstraction layer. It uses a tool-use architecture where the AI agent can invoke file operations, terminal commands, and search tools. The extension maintains conversation context and project awareness through file indexing and intelligent context selection. Each provider integration follows a unified interface, making it straightforward to add new model backends.

Self-Hosting & Configuration

  • Install from the VS Code marketplace or JetBrains plugin repository
  • Configure your preferred LLM provider and API key in the extension settings
  • Supports OpenAI-compatible API endpoints for self-hosted models
  • Custom system prompts and agent behaviors are configurable via settings
  • Works with any project type and language supported by the IDE

Key Features

  • Provider-agnostic design supporting 10+ LLM providers out of the box
  • Full codebase awareness with multi-file editing capabilities
  • Built-in terminal command execution for running builds and tests
  • Open-source with an active community contributing features and fixes
  • Available for both VS Code and JetBrains IDE families

Comparison with Similar Tools

  • Cursor — proprietary IDE fork; Kilocode is an open-source extension for existing IDEs
  • Continue — similar open-source agent but Kilocode emphasizes multi-IDE support
  • GitHub Copilot — subscription-based autocomplete; Kilocode is free and supports agentic workflows
  • Cline — VS Code-only agent; Kilocode also supports JetBrains
  • Aider — terminal-based agent; Kilocode provides a graphical IDE experience

FAQ

Q: Is Kilocode free to use? A: Yes. Kilocode is open-source and free. You provide your own LLM API keys.

Q: Can I use local models with Kilocode? A: Yes. It supports Ollama, LM Studio, and any OpenAI-compatible local endpoint.

Q: Does Kilocode work with JetBrains IDEs? A: Yes. It supports IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains products.

Q: How does Kilocode handle large codebases? A: It uses intelligent context selection to include only relevant files in the LLM prompt, keeping token usage manageable.

Sources

讨论

登录后参与讨论。
还没有评论,来写第一条吧。

相关资产