Fix Env

What it does

When your agent uses it

• A tools setup.sh fails
• An env breaks after a system update
• A standalone venv needs a fix — for your own use

Example prompts

• “Use the fix-env skill to fix tool environment setup failures in proto-tools, either just for the current machine (eject the tool's standalone dir…”
• “/fix-env”

Requirements

• Python 3
• Pre-approved tools (allowed-tools): Read, Write, Bash, Glob, Grep

Workflow steps

7 steps, taken from the step headings in SKILL.md.

1. 1Read STATUS.txt
2. 2Verify Compute Detection (Pre-Flight)
3. 3Verify Env Isolation (Pre-Flight)
4. 4Match Error to Pattern
5. 5Apply the Fix to setup.sh
6. 6Validate the Fix
7. 7Document What You Changed

What it can do on your machine

Read from SKILL.md and the folder at commit 64363bd. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves these tools, so the agent can use them without asking each time:

  • Read
  • Write
  • Bash
  • Glob
  • Grep

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • pytest
  • python
  • pip
  • uv

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md. Its commands use pip and uv, which can reach the network depending on how they are called.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Fix Env loads about 2.5k tokens when it runs. Until then it costs about 224 tokens; SKILL.md has 1,032 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~224

When it runs · the whole SKILL.md, loaded when a task matches

~2.5k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

• NotePre-approves every shell command (allowed-tools: Bash)SKILL.md

  allowed-tools: Read, Write, Bash, Glob, Grep

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from evo-design/proto-tools at commit 64363bd, republished under its MIT licence (© evo-design). 1,032 words, ~2,550 tokens.

name

fix-env

description

Fixes tool environment setup failures in proto-tools, either just for the current machine (eject the tool's standalone dir, patch it, and point PROTO_<TOOLKIT>_STANDALONE_DIR at it; works for any install, including a non-editable pip install) or as a cross-platform fix contributed back to the repo. Same diagnosis for both: infrastructure failures (compute detection, env variable isolation, sitecustomize.py injection, micromamba install), PyTorch/CUDA issues (ABI mismatch, broken symlinks, triton coordination), JAX setup/downgrades, compilation failures (GCC/nvcc mismatch, source builds), network/binary download failures, platform issues (aarch64, Python version), and device management setup failures (standalone helpers, CUDA visibility). Use when a tool's setup.sh fails, an env breaks after a system update, or a standalone venv needs a fix — for your own use or to upstream.

allowed-tools

Read, Write, Bash, Glob, Grep

cat tool_envs/{tool}_env/STATUS.txt

# Check actual hardware
nvidia-smi

# Check what the detection set
python -c "
from proto_tools.utils.compute_deps import detect_compute_environment
env = detect_compute_environment()
for k, v in sorted(env.items()):
    print(f'{k}={v}')
"

# Inside a tool subprocess, check critical env vars:
# CONDA_PREFIX and VIRTUAL_ENV should point to tool_envs/{tool}_env
# LD_LIBRARY_PATH should include cuda_env/lib for CUDA JIT tools
# python sys.prefix should match tool env path

rm -rf tool_envs/{tool}_env
pytest -k "tool_key" --all -sv
pytest --cpu-only --skip-ci
pytest --gpu-only --all  # if GPU available

© evo-design, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in .claude/skills/fix-env of evo-design/proto-tools.

• SKILL.md
• PATTERNS.md

Open the folder on GitHubat commit 64363bd