GitHub Repository

Agent-SDK without CLI dependencies, as an alternative to claude-agent-sdk, completely open source

2,717 starsTypeScript

Open-agent-SDK – Claude Code's internals, extracted and open-sourced

Name: Open-agent-SDK – Claude Code's internals, extracted and open-sourced
Availability: InStock
Author: idoubi

by idoubi·Apr 2, 2026·7 points·1 comment

Visit Project View on HN

AI Analysis

●●SolidBold BetWizardry

In-process agent loop eliminates CLI overhead, but relies on leaked proprietary source maps.

Strengths

•Pure function calls remove subprocess overhead, enabling thousands of concurrent agents.
•Drop-in compatibility with claude-agent-sdk interface minimizes migration friction for existing projects.
•Supports custom MCP servers and Zod schemas for flexible tool integration.

Weaknesses

•High legal risk; Anthropic could issue cease-and-desist anytime given the source origin.
•Maintenance depends on reverse-engineering future Claude Code updates rather than official APIs.

Post Description

Claude Code's source map leaked last month. Instead of just reading it, I fed the entire source into Claude Code itself and asked it to extract the core agent loop logic.

The result: open-agent-sdk — a drop-in replacement for claude-agent-sdk that's fully open source and doesn't spawn a CLI subprocess.

Why this matters if you've built with claude-agent-sdk:

claude-agent-sdk is just a thin wrapper around the Claude Code binary. It works, but it's a black box — when something breaks, you're stuck.

Every query creates a new Claude Code process. That's fine on a laptop, not fine when you're running thousands of concurrent agents in the cloud.

What open-agent-sdk does differently:

Pure function calls, no CLI process spawning — cloud-native from day one Fully compatible interface with claude-agent-sdk — swap the package name, done

MIT licensed — fork it, patch it, make it yours

→ https://github.com/codeany-ai/open-agent-sdk-typescript