vllm-ascend-workspace

Add, verify, repair, or remove a managed remote NPU host for this workspace. Use for requests like “配置服务器”, “加一台机器”, “检查 ready”, “修容器 SSH”, or “移除机器”. Do not use for code sync, rebuilds, serving, or benchmarking.

Collect one Ascend torch-profiler case end-to-end on a workspace-managed remote NPU container. Starts a profiled vLLM service, brackets a workload with /start_profile and /stop_profile, runs analyse(), verifies kernel_details.csv landed, and writes a manifest the analysis skill can consume. Use for requests like "采集 profiling", "torch profiler 跑一个 case", "采一份 profile 出来", "采 profiling 给我分析". Do not use for pure performance benchmarking, HBM/memory profiling, or for analysing already-collected profiling data (that is the analysis skill's job).

Ensure a ready remote runtime runs the exact current local workspace state before any remote smoke, service launch, or benchmark. Use automatically immediately before remote execution when direct local -> container SSH already works and local uncommitted changes must be reflected remotely. Do not use for initial machine attach, generic Git topology work, or unrelated local-only coding.

Initialize this workspace after clone. Use for requests like “初始化仓库”, “配置 gh / GitHub 登录”, “初始化子模块”, or “把 vllm / vllm-ascend remotes 改成我的 fork”. Do not use for ordinary coding, serving, benchmarking, or unrelated Git tasks.

Start, check, or stop a single-node vLLM Ascend online service on a workspace-managed ready remote container. Use for requests like "拉服务", "在远端起个服务", "重启服务", "看服务状态", "停掉服务". Do not use for machine attach, environment bootstrap, code sync, benchmark orchestration, or offline inference.

Download, resume, status-check, and SHA256-verify ModelScope model weights. Use for $modelscope download/status/verify/check, Chinese requests to 下载/续传/补全/查看进度/校验 ModelScope 权重, and tasks that need durable background ModelScope downloads under explicit local directories.

Analyze Ascend NPU torch profiler output (kernel_details.csv / trace_view.json / op_summary / communication.json) for one or many profiling roots and produce a traceable report (rank/step/layer/operator summary, cross-rank alignment, diagnosis findings, report.md / report.xlsx / report.html with single-step inspectors, bubble tracing axes, and zoomable Chrome-tracing-style timelines). Use for requests like "分析 profiling", "解析这份 kernel_details", "看 step/layer 切分", "跨 rank 对齐", "通信慢/EP 不均/快慢卡", "生成 profiling 报告". Do not use for HBM/显存归因 (use ascend-memory-profiling), service lifecycle (use vllm-ascend-serving), benchmarks (use vllm-ascend-benchmark), or采集 profiling 数据 (use ascend-profiling-collection).

Profile and attribute HBM memory usage on Ascend NPU for vLLM serving scenarios. Breaks down memory into fixed overhead, model weights, KV cache, HCCL buffers, activations, and runtime, with traceable evidence chains. Use for requests like "分析显存占用", "显存 profiling", "HBM 用了多少", "内存各部分拆分". Do not use for performance profiling (kernel timing, throughput), offline inference, or non-Ascend hardware.

Resolve, probe, execute, observe jobs, sync, manage service lifecycle, transfer artifacts, and clean VAWS remote Ascend session containers through structured agent-facing CLI entrypoints.

Create, list, inspect, remove, and garbage-collect isolated VAWS agent sessions. Use before remote execution when multiple agent tasks must run in parallel without sharing local worktrees, remote containers, serving state, or resource leases.

Run vLLM online-serving benchmarks on a workspace-managed remote container. Use for requests like "跑个 benchmark", "对比性能", "压测一下", "测下吞吐", or "看下有没有性能回退". Do not use for accuracy tests, nightly CI matrix runs, offline inference, or service-only lifecycle.

Start, check, or stop a single-node vLLM Ascend online service on a workspace-managed ready remote container. Triggered when users ask to launch, restart, check status, or stop a vLLM Ascend service. Do not use for machine attach, environment bootstrap, code sync, benchmark orchestration, or offline inference.

Add, verify, repair, or remove a managed remote NPU host for this workspace. Use for requests like "configure server", "add a machine", "check ready", "fix container SSH", or "remove machine". Do not use for code sync, rebuilds, serving, or benchmarking.

Ensure the remote runtime runs the exact current local workspace state before any remote smoke test, service launch, or benchmark. Triggered automatically before remote execution when direct local-to-container SSH works. Do not use for initial machine attach, generic Git topology work, or unrelated local tasks.

Initialize the workspace after clone. Use for requests like "init repo", "configure gh / GitHub login", "init submodules", or "set up forks / remotes".