</>MCP Agents Market
Agent

claude-code

by claude-code-best22.1kTypeScriptUpdated 2026-08-18

原汁原昧 Claude Code 可运行,可构建, 可调试版; 生产级工程化, 企业级可靠性; 安全无毒, 内存泄露修复

Claude CodeZedCursor

Claude Code Best (CCB) is an enterprise-grade AI agent that fully replicates and extends Anthropic's official Claude Code implementation with production-ready engineering and enhanced capabilities. It provides autonomous goal-driven task execution, multi-agent orchestration through Ultracode workflows, and enterprise monitoring via Langfuse and Sentry. The agent supports browser automation, computer use, voice input, web search, and integrates with IDEs through ACP protocol compatibility with Zed and Cursor.

Key Features

Goal-driven autonomous execution with `/goal` command, automatic cross-turn task continuation, pause/resume support, and network interruption handling
Ultracode multi-agent orchestration with deterministic JavaScript workflow scripts supporting agent/pipeline/parallel/phase patterns and concurrent execution caps
Artifacts system for uploading HTML dashboards and reports to public URLs with Cloudflare Worker + R2 self-hosting support
Enterprise monitoring with Langfuse for agent loop visibility, Sentry error tracking, and GrowthBook feature flags
Pipe IPC for multi-instance collaboration enabling same-machine and LAN cross-machine zero-config agent discovery and communication
ACP protocol support for IDE integration with session resumption and Skills permission bridging in Zed and Cursor
Built-in web search (Bing/Brave), browser automation (Chrome Use), computer control, and voice input with Doubao ASR
Poor Mode for reduced token usage, custom model provider support (OpenAI/Anthropic/Gemini/Grok), and remote control self-hosting via Docker

Use Cases

  • 01Running long-running autonomous development tasks with goal-based execution that persists across sessions and network interruptions
  • 02Orchestrating multiple AI agents in parallel or sequential workflows for complex multi-step automation tasks
  • 03Monitoring production AI agent behavior with enterprise-grade observability through Langfuse traces and Sentry error tracking
  • 04Integrating AI assistance directly into IDEs like Zed and Cursor through ACP protocol with session persistence
  • 05Building self-hostable AI agent deployments with Docker-based remote control accessible from mobile devices
  • 06Creating interactive data visualizations and reports that agents automatically upload to shareable public URLs

Related Agents

View more

claude-code — FAQ

What is Claude Code Best (CCB)?+

Claude Code Best is an open-source AI agent that recreates Anthropic's Claude Code with enterprise-grade engineering, adding goal-driven execution, multi-agent orchestration, enterprise monitoring, and features only available in Claude's paid tiers. It supports custom model providers compatible with Anthropic, OpenAI, Gemini, and Grok APIs.

How do I install Claude Code Best?+

Install globally via npm with `npm i -g claude-code-best`, then run `ccb` to start. Alternatively, clone the repository and run `bun install` followed by `bun run dev` for development mode. Bun version 1.3.11 or higher is required for source installation.

Do I need an Anthropic API key to use this agent?+

No, CCB supports any Anthropic-compatible API provider. After first run, use `/login` and select 'Anthropic Compatible' to configure third-party API services with your base URL and API key. OpenAI and Gemini protocols are also supported.

Which AI clients and IDEs work with Claude Code Best?+

CCB supports ACP protocol integration with Zed and Cursor IDEs. It can also run standalone in terminal mode or via self-hosted remote control interface accessible from any device including mobile phones.

Is Claude Code Best free to use?+

Yes, the agent software is open-source and free. However, you will need API access to a compatible language model provider, which may have its own costs depending on the service you choose.

What are the main differences from official Claude Code?+

CCB adds enterprise features like Langfuse monitoring, Sentry error tracking, goal-driven autonomous execution, multi-agent Ultracode workflows, Pipe IPC for agent collaboration, self-hosted Artifacts, and Poor Mode for cost reduction. It also includes features from Claude's enterprise tier without requiring account login.

How do I install claude-code?+

Open the source repository on GitHub and follow its README. claude-code is a agent — MCP Agents Market links you directly to the official repo.

Is claude-code free?+

claude-code is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches