Hep Agent Sandbox — AI skill for Claude Code
Sandboxed VS Code dev container for AI coding agents doing HEP work on a Mac (Claude Code plugin + manual guide).
How to install Hep Agent Sandbox
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open S81D/hep-agent-sandbox and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Hep Agent Sandbox does
Sandboxed VS Code dev container for AI coding agents doing HEP work on a Mac (Claude Code plugin + manual guide).
Alternatives in AI
- Coding Agent Template — Multi-agent AI coding platform powered by Vercel Sandbox and AI Gateway 1.8k ★
- Open Dot — Open-source personal AI agents that work on their own, on their own computers 606 ★
- Late CLI — High-performance AI agent for long-horizon tasks and local models 442 ★
README
hep-agent-sandbox
A sandboxed VS Code dev container for AI coding agents (Claude Code, Codex) doing high-energy-physics work on a Mac. Agents run inside a Linux container that sees **only your research folder** (default `~/hep`), with the tools HEP work needs, instead of running directly on your Mac with access to everything.
**What you get**
- HEP tooling: Python 3.14 with ROOT 6.40, uproot, awkward, hist, iminuit, pyhepmc, jupyterlab (pinned via conda-forge); gcc/clang/cmake/gdb; a TeX Live subset (REVTeX, biber, LuaLaTeX, XeLaTeX); poppler for PDFs; Node.js; the Claude Code and Codex CLIs.
- Isolation: one bind mount of your research folder, with optional read-only dataset folders. No sudo, all Linux capabilities dropped, no Docker socket, no Mac credentials.
- GitHub over SSH: through your Mac's forwarded ssh-agent; your key never enters the container.
- Optional remote-access presets: none, Fermilab via Kerberos (host-only), or SSH-key clusters.
- A verification script: checks the toolchain and the isolation from inside the container.
- An agent policy: instructions every agent session loads: scientific safety, provenance, grid stewardship.
- An agent survey to improve the setup from your agents' actual experience.
Read [docs/DESIGN.md](docs/DESIGN.md) for the security model: what is enforced, what is only an instruction, and the risks that remain.
Quick start with Claude Code (guided)
You need Docker Desktop, VS Code with the Dev Containers extension, and Claude Code on the Mac; see the prerequisites linked below. In Claude Code on the Mac:
/plugin marketplace add S81D/hep-agent-sandbox
/plugin install hep-sandbox@hep-agent-sandbox
Then ask: *"Set up the HEP agent sandbox"* (or run `/hep-sandbox:setup-sandbox`).
The skill takes you through these steps, asking before anything that changes your Mac:
- a read-only check of the Mac
- a few decisions
- generating the configuration
- building and veri
Related Skills
Orbstack Sandbox
Sandboxed OrbStack VMs for LLM coding agents — locks down host filesystem access so agents in permissive modes
Dozer Sandbox
Fast, resumable, secure Linux sandboxes for AI coding agents on your Mac — your API keys never go inside.
Devsandbox
Run AI coding agents in a per-project sandbox that keeps your real dev environment. Your shell, mise-managed t
Machogs
Your Mac is slow, hot and loud. machogs tells you what is doing it, in plain English. Finds the AI tool helper
Claude Buddy Mac
Pixel desktop pet for macOS that shows what Claude is doing across Claude Code, Cowork and claude.ai — built w
Domain Name Brainstormer
Generates creative domain name ideas for your project and checks availability across multiple TLDs (.com, .io,
Related Agents
Tsk - AI Agent Task Manager And Sandbox
by dtormoen - A Rust CLI tool that lets you delegate development tasks to AI agents running in sandboxed Docke
Oneshot CI Fixer
Fix sandbox agent. Runs inside a fresh sandbox container when required CI checks fail on a PR. Reads the failu
Swe Dev Loop Systems Architect
Designs and audits coding-agent loops, debugging and migration workflows, CI/release loops, autonomous-science