Sandcastle

Active
GitHub TypeScript MIT

Description

A TypeScript tool for orchestrating sandboxed coding agents with secure execution environments powered by sandcastle.run.

Key Features

  • Sandboxed agent orchestration — safely execute AI coding agents in isolated sandboxes via sandcastle.run()
  • Multiple sandbox providers — built-in Docker, Podman, Vercel support, custom sandbox creation
  • Provider-agnostic design — swappable agent providers, built-in Claude Code support
  • Git branch strategy management — automatic agent branch creation, commits, and merges
  • Parallel agent execution — supports multiple AFK agents running in parallel, review pipelines
  • TypeScript native — full TypeScript support with type-safe API design

Use Cases

💡 Parallel AI coding — run multiple Claude Code agents in isolated sandboxes for different tasks
💡 Code review pipelines — create automated agent code review and merge workflows
💡 Secure code execution — safely execute untrusted AI-generated code in containerized environments
💡 CI/CD agent integration — use sandboxed agents in continuous integration pipelines
💡 Multi-repo parallel development — run agents across multiple workspaces for cross-repo modifications

Tags

coding-agent sandbox typescript orchestration

Categories

💻 Coding Agent 🏗️ Sandbox & Execution

Quick Start

# Install
npm install --save-dev @ai-hero/sandcastle

# Initialize project
npx @ai-hero/sandcastle init

# Configure .sandcastle/.env
# Set ANTHROPIC_API_KEY

# Run the agent
npx tsx .sandcastle/main.ts

# Or via API
import { run, claudeCode } from "@ai-hero/sandcastle";
import { docker } from "@ai-hero/sandcastle/sandboxes/docker";

await run({
  agent: claudeCode("claude-opus-4-7"),
  sandbox: docker(),
  promptFile: ".sandcastle/prompt.md",
});
Visit GitHub

Related Projects