Stats
Actions
Available In
Tags
Plugin
gitcode-publish
Publish model README to GitCode with auto-inferred YAML frontmatter tags from HuggingFace/ModelScope API and local transformers config.
README
g-claude
A Claude Code skills marketplace for Ascend NPU inference, model quantization, LLM evaluation, and structured Git commits.
Install as a marketplace
claude plugin marketplace add starmountain1997/g-claude
Then install individual skills:
claude plugin install ascend@g-claude
claude plugin install vllm-ascend@g-claude
claude plugin install msmodelslim@g-claude
claude plugin install aisbench@g-claude
claude plugin install commit-as-prompt@g-claude
Or install everything at once (also installs karpathy-skills and skill-creator):
curl -fsSL https://raw.githubusercontent.com/starmountain1997/g-claude/main/scripts/install-g-claude.py | python3 - --update
Pass any script arguments after python3 -, e.g. --update, --help, etc.
Install via OpenPackage
OpenPackage is a universal package manager for coding agent configs. Install all skills at once:
npm install -g opkg
opkg install gh@starmountain1997/g-claude
Or install individual skills:
opkg install gh@starmountain1997/g-claude --skills ascend
opkg install gh@starmountain1997/g-claude --skills vllm-ascend
opkg install gh@starmountain1997/g-claude --skills msmodelslim
opkg install gh@starmountain1997/g-claude --skills aisbench
opkg install gh@starmountain1997/g-claude --skills commit-as-prompt
Skills
| Skill | Description |
|---|---|
| ascend | Ascend NPU hardware entry point — health check, environment setup, shell script template. Starting point for any Ascend workflow. |
| vllm-ascend | vLLM-Ascend serving toolchain — install, offline validation, scenario tuning, online serving, contribution guide. |
| msmodelslim | Model quantization on Ascend NPUs — W4A8/W8A8/W4A4, one-click and custom YAML, MoE mixed precision, VLM support, accuracy recovery. |
| aisbench | AISBench evaluation framework — accuracy benchmarks (GSM8K, MMLU, AIME) and performance benchmarks against vLLM services. |
| commit-as-prompt | Structured Git commits with WHAT/WHY/HOW format, optimized as AI context for future sessions. |
Workflow
The Ascend skills form a pipeline:
ascend (NPU check)
├──► vllm-ascend (install → serve)
├──► msmodelslim (quantize → serve via vllm)
└──► aisbench (evaluate accuracy & performance)
License
MIT
Stats
Versionv0.0.11
LanguagePython
Stars0
MaintenanceExcellent
Last Commit
Added
Available In
Tags
Categories
