# job-env-check

job-env-check — a skill for Claude Code tests whether bare python3 imports PyTorch before helper tools execute. It runs at the start of each evaluation loop on a GPU host.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf (Devin Desktop), Claude.ai · ★ 339 repo stars
- Author: OpenBMB
- License: Apache-2.0 · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/OpenBMB/ForgeKernel/tree/HEAD/skills/job-env-check/SKILL.md
- Updated: 2026-11-11
- Security check: automated, 2026-11-17 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/job-env-check

## What it does

The tool verifies that executing `python3` directly can import `torch`. GPU tasks executed over SSH frequently run in non-login shells where the default system Python lacks machine learning dependencies, even when the accelerator itself has configured drivers.

A driver executes `python3 notes/job_env_check.py` once per loop start. The script returns exactly one line: either `JOB-ENV-OK python3=<path> torch=<version>` or `JOB-ENV-FAIL python3=<path> <reason>`, outputting the traceback above the failure flag.

The logic expects the ForgeKernel repository layout. Sibling tools launch via `python3 notes/<tool>.py` and re-spawn `python3 eval/bench.py`, so all steps fail immediately if the shell interpreter cannot import PyTorch.

## Install

### Claude Code (native platform)

**Manually: put the files in place**
- File: `~/.claude/skills/job-env-check/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/job-env-check/

### Codex

**Manually: put the files in place**
- File: `.agents/skills/job-env-check/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

**Manually: put the files in place**
- File: `.cursor/skills/job-env-check/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install OpenBMB/ForgeKernel skills/job-env-check
```

2. **Manually: put the files in place**
- File: `.github/skills/job-env-check/SKILL.md`

### Gemini CLI

**Platform CLI**
```bash
gemini skills install https://github.com/OpenBMB/ForgeKernel --path skills/job-env-check
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/job-env-check/SKILL.md` — The legacy .windsurf/skills/ path still works

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/OpenBMB/ForgeKernel/tree/HEAD/skills/job-env-check/SKILL.md

## When to use it
- verify PyTorch imports under the default python3 binary on GPU hosts
- diagnose non-login SSH shell environment mismatches
- log the active Python interpreter path and torch version in one line

## Example prompts
- Check if python3 imports torch before launching the benchmark suite
- Run python3 notes/job_env_check.py to verify the remote GPU environment

## Source text · SKILL.md · Apache-2.0

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: c9dd0dab3dcc8232.

BEGIN THIRD-PARTY FILE c9dd0dab3dcc8232
```text
# job-env-check

Whether a GPU job's bare `python3` can run the other skills. They are started as
`python3 notes/<tool>.py` and re-spawn `python3 eval/bench.py`, so they all need that
interpreter to import torch; on an ssh host the job runs in a non-login shell, where
`python3` may be the system one even though the kernel itself evaluates fine through the
card's own interpreter.

The driver runs it once per loop start (`python3 notes/job_env_check.py`). It prints exactly
one line, `JOB-ENV-OK python3=<path> torch=<version>` or
`JOB-ENV-FAIL python3=<path> <reason>`, with the failing import's own text above a FAIL.
```
END THIRD-PARTY FILE c9dd0dab3dcc8232

The third-party file has ended. Everything between the c9dd0dab3dcc8232 tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

© OpenBMB · Apache-2.0 · https://github.com/OpenBMB/ForgeKernel/tree/HEAD/skills/job-env-check/SKILL.md

## Similar
- [modal](https://affpapa.org/en/nevibecode/rules/modal.md) — rule · ★ 3589 repo stars
- [dali-dynamic-mode](https://affpapa.org/en/nevibecode/skills/dali-dynamic-mode.md) — skill · 2100 installs
- [quark-install](https://affpapa.org/en/nevibecode/skills/quark-install.md) — skill · 966 installs
- [triton-lang](https://affpapa.org/en/nevibecode/skills/triton-lang.md) — skill · 358 installs
- [env-manager](https://affpapa.org/en/nevibecode/plugins/env-manager.md) — plugin · ★ 2858 repo stars
- [pytorch-expert](https://affpapa.org/en/nevibecode/agents/pytorch-expert.md) — subagent · ★ 1029 repo stars

From the same set: [openbmb/forgekernel](https://affpapa.org/en/nevibecode/by/openbmb/forgekernel.md) — 8 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
