sandcastle SKILL.md skillA

sandcastle from av/skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add av/skills/sandcastle@git:20260519.b676141

Pins git:20260519.b676141 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
av/skills
Path
sandcastle/SKILL.md
Versions recorded
1
Latest version
git:20260519.b676141, committed May 19, 2026
Size
14,499 bytes
Repository stars
28
License
MIT

Orchestrate AI coding agents (Claude Code, Codex, OpenCode) in isolated sandboxes using the @ai-hero/sandcastle SDK. Use when the user needs to (1) run agents AFK in Docker/Podman containers, (2) build multi-agent pipelines with plan-execute-review patterns, (3) run parallel agents on separate worktrees, (4) create iterative agent loops with maxIterations, (5) extract structured output from agent runs, (6) set up sandcastle in a new or existing project, or (7) write prompt files with template…

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# Sandcastle

Sandcastle (`@ai-hero/sandcastle`) orchestrates AI coding agents inside isolated sandbox environments. It manages git worktrees, boots containers, runs agents with structured prompts, collects commits, and merges results back to branches — all from a single `run()` call.

## Installation

```bash
npm i @ai-hero/sandcastle
npx sandcastle init          # scaffolds .sandcastle/ with Dockerfile and prompt templates
npx sandcastle docker build-image   # builds the Docker image from .sandcastle/Dockerfile
```

## Quick Start

```typescript
import { run, claudeCode } from "@ai-hero/sandcastle";
import { docker } from "@ai-hero/sandcastle/sandboxes/docker";

const result = await run({
  agent: claudeCode(),
  sandbox: docker(),
  promptFile: "./.sandcastle/prompt.md",
});

console.log(result.commits);   // [{ sha: "abc123" }, ...]
console.log(result.branch);    // branch the agent worked on
```

## Core Concepts

### The Four Entry Points

| Function | Use Case |
|---|---|
| `run(options)` | One-shot agent invocation. Full lifecycle managed automatically. |
| `createSandbox({ branch, sandbox })` | Reusable sandbox on an explicit branch. Call `.run()` multiple times. |
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add av/skills/sandcastle@sha256:5f5719f7e481b4a7

The line at the top of this page pins the label git:20260519.b676141, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_3ibchy52jc42miw3.svg)](https://markdownregistry.com/a/art_3ibchy52jc42miw3)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 1 version of this file, committed on May 19, 2026.

versioncommittedcommitsizeaudit
git:20260519.b676141 latest2026-05-19 b676141 14,499 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (14499 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

av/skills · 28 stars · license MIT · pushed 2026-10-21 · branch master

API

GET https://markdownregistry.com/api/v1/artifacts/art_3ibchy52jc42miw3
GET https://markdownregistry.com/api/v1/resolve?ref=av/skills/sandcastle
GET https://markdownregistry.com/api/v1/blob/5f5719f7e481b4a76fc8125862933eed35050fdf6fe8f08b623b07d768244acf

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from av/skills

agent-integration-testing SKILL.md skill
av/skills · agent-integration-testing/SKILL.md · Use when the user requests integration testing, feature validation, or test plan execution
git:20260330.d629481 · audit A · 28 stars
anneal SKILL.md skill
av/skills · anneal/SKILL.md · Use when the user wants to systematically fix AI code slop — duplicated logic, over-engineering, silent error…
git:20260601.cc3497a · audit A · 28 stars
article-factory SKILL.md skill
av/skills · article-factory/SKILL.md · Produce a researched long-form article from a topic prompt via an orchestrated pipeline - research agent (first-person…
git:20260730.ada63fd · audit A · 28 stars
autoresearch SKILL.md skill
av/skills · autoresearch/SKILL.md · Runs autonomous keep/discard experiments on a codebase to optimize a single metric for a fixed duration, in the style…
git:20260923.7fb0398 · audit A · 28 stars
boost-modules SKILL.md skill
av/skills · boost-modules/SKILL.md · Create custom modules for [Harbor Boost](https://github.com/av/harbor/tree/main/boost), an optimizing LLM proxy. Use…
git:20260123.eeaa581 · audit B · 28 stars
bugbash SKILL.md skill
av/skills · bugbash/SKILL.md · Systematically explore and test any software project (CLI, API, Backend, Library, etc.) to find bugs, usability issues…
git:20260402.6abcf0b · audit A · 28 stars
bughunt SKILL.md skill
av/skills · bughunt/SKILL.md · Fully autonomous bug hunting pipeline — discover bugs in a scoped area using parallel subagents, independently triage…
git:20260612.7c5d841 · audit A · 28 stars
catalog-deslop SKILL.md skill
av/skills · catalog-deslop/SKILL.md · Catalog-then-fix slop removal on an existing prose draft. Parallel catalog subagents inventory four slop categories…
git:20260730.ada63fd · audit A · 28 stars
discipline SKILL.md skill
av/skills · discipline/SKILL.md · Bulletproof agent operating protocol. 15 failure-prevention rules distilled from 120+ real sessions and 10 agent…
git:20260509.2572fa1 · audit A · 28 stars
factual-communication SKILL.md skill
av/skills · factual-communication/SKILL.md · Restate content as a flat list of atomic facts in the style of av/facts fact sheets. Use when the user says "just…
git:20260825.4dce9e4 · audit A · 28 stars
ideate SKILL.md skill
av/skills · ideate/SKILL.md · Timeboxed ideation on a topic where lens-driven proposers generate structurally different ideas, blind critics cull…
git:20261011.7d38584 · audit A · 28 stars
improve-skill SKILL.md skill
av/skills · improve-skill/SKILL.md · Improve an existing agent SKILL.md through one gated cycle: select the target (default this skill), snapshot the…
git:20260813.b71baeb · audit A · 28 stars

Every file in av/skills

Browse by kind, by grade A, or by owner.