菜单
Develop HyperParallel platform abstraction: new Platform APIs, torch/mindspore backends, FSDP/HSDP/Pipeline/Activation Checkpoint, DTensorBase, collectives. Use when changing platform/ or cross-backend features. Not for shard op YAML/impl (use dist-op-dev) or git/CI (autogit / gate-doctor).
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/mindspore-ai/hyper-parallel --skill platform-dev命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Add or extend HyperParallel unit tests under tests/ut. Use when writing UT, increasing coverage, or mocking distributed/hardware. Not for multi-card ST (see testing.md / distributed-op-testing) or shard-op-specific UT constraints (distributed-op-testing rule).
GitCode fork workflow: commit, push, create/append PR, status, squash, lint/test. Triggers: 帮我提交, create PR, PR 状态, /commit, /create-pr, etc. Origin=fork, upstream=main. Details in references/.
Review HyperParallel changes for distributed correctness, stream sync, memory safety, cross-platform parity, and code quality. Use for /code-review, PR review, or when the user asks to review/check changes.
基于 SOC 职业分类
正在显示 SKILL.md
| name | platform-dev |
| description | Develop HyperParallel platform abstraction: new Platform APIs, torch/mindspore backends, FSDP/HSDP/Pipeline/Activation Checkpoint, DTensorBase, collectives. Use when changing platform/ or cross-backend features. Not for shard op YAML/impl (use dist-op-dev) or git/CI (autogit / gate-doctor). |
Guides cross-platform work under platform/. This file is the index —
load the matching workflow / reference before that step.
platform/platform.py, or impl in platform/torch|mindspore/Tree, API map, patterns: references/architecture.md, references/quick-reference.md, references/decisions.md.
import torch / mindspore in platform-agnostic code — get_platform()platform/platform.py)NotImplementedErrordevice/ProcessGroup vs mindspore str groups — check signaturesplatform/torch|mindspore/ + # pylint: disable=C0415handle.wait(), event.record→wait, resize_(0) — see .agent/rules/distributed.mdparallel_*.py → dist-op-dev