vllm-ascend-workspace: Skill for Claude Code

.agents/skills/ascend-profiling-analysis/SKILL.md

ascend-profiling-analysis is a skill for Claude Code, Codex from maoxx241/vllm-ascend-workspace. It costs 191 tokens per session (9,111 once invoked), scanned A, original, MIT.

A report generator for Ascend NPU PyTorch profiler files, which record how AI workloads run on Huawei Ascend processors. It summarizes steps, layers, operators, ranks, and communication across one or more profiling runs.

In plain words
What is it for?
Use it to inspect step, layer, and operator timing; compare worker ranks; investigate pipeline gaps, CPU work, waiting points, slow devices, and unbalanced expert-parallel workloads; and produce Markdown, Excel, or HTML reports.
Why use it?
It turns large, hard-to-read profiling files into findings that can be traced back to the original data. This helps investigate slow communication, uneven workload distribution, idle time, and differences between devices.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: installed under .agents/ (shared by several agents).

This is maoxx241/vllm-ascend-workspace's own configuration. It tells Claude Code and Codex how to work on vllm-ascend-workspace itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything vllm-ascend-workspace configures →

Reuse

Borrowing it

Nothing to install: this file belongs to maoxx241/vllm-ascend-workspace. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/maoxx241/vllm-ascend-workspace/main/.agents/skills/ascend-profiling-analysis/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/maoxx241/vllm-ascend-workspace

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for ascend-profiling-analysis

README.md
[![agentmods](https://agentmods.dev/badge/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis/github.svg)](https://agentmods.dev/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis)
Your own site
<a href="https://agentmods.dev/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis"><img src="https://agentmods.dev/badge/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for ascend-profiling-analysis

Your own site · 80×15
<a href="https://agentmods.dev/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis"><img src="https://agentmods.dev/badge/skills/maoxx241/vllm-ascend-workspace/ascend-profiling-analysis.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 191 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 9,111 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 1 finding, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium analysis-evasion · line 1
    Suspicious Unicode normalization or mixed-script content
    Fix: Review the flagged content for security risks. Ensure no credentials, secrets, or sensitive data are exposed.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00191 $0.09111
Opus 5 $0.00096 $0.04555
Sonnet 5 $0.00038 $0.01822
Haiku 4.5 $0.00019 $0.00911

Measured 8d ago against content hash 0530cf2b982c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

ascend-profiling-analysis scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

The scan reads SKILL.md. This mod also ships 34 executable files (scripts/_common.py, scripts/ascend_profile/__init__.py, scripts/ascend_profile/analyze.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/ascend-profiling-analysis/SKILL.md · 401 lines

How it starts

The opening of the file, as written. The whole thing — 401 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Ascend Profiling Analysis

Status: experimental / beta. 当前 PR 主要提供:远端 pipeline、evidence-chained report、HTML 三级聚焦视图、stage selector。Knowledge 已分层:kernel 分类规则(kernel_signatures.yaml:match_rules)、attention 家族判定(attention_families.yaml:cheat_sheet.resolver)、diagnosis 阈值与文案(diagnosis_rules.yaml)、segment 层锚点先验(segmentation_rules.yaml)均为运行时加载的 YAML;仍在 Python 内的是 segment.py 切分策略、classify.py block 拆分、以及 finding 的触发条件(见 Knowledge map)。新模型 / 新算子族碰到问题时,优先改 knowledge YAML;改 Python 前请把 counterexample 落到 knowledge/known_counterexamples.md

Remote substrate rule: use .remote-dev remote tools for ad hoc remote read/edit/bash/search/patch work around profiling roots and generated reports. Use this skill for the domain analysis workflow and keep its scripts as the compatibility backend for managed VAWS sessions.

读取 Ascend NPU torch profiler 的产物 (kernel_details.csv, trace_view.json, op_summary, communication.json 等),做 normalize → segment → summarize → cross-rank → diagnostics → report 的端到端分析,产物全部可追溯到原始 row range。

本 skill 只消费已经采集好的 profiling root,不负责采集,不负责服务生命周期,不负责 benchmark。

Use this skill when

  • 用户提供一个 profiling root 路径(远端或工作区路径),或者 ascend-profiling-collection 写出的 manifest.json,要求分析。
  • 用户问 step / layer / operator 统计、跨 rank 对齐、bubble、AICPU、wait anchor。
  • 用户怀疑通信慢、EP 负载不均、快慢卡、陪跑/dummy rank、workload 非对称。
  • 用户需要带 evidence 链的 report.md / report.xlsx / report.html(HTML 报告是单文件零依赖,含交互式 Single-step Inspector、bubble tracing axis、可缩放多流时间轴、46 字段算子卡)。
  • 用户要在多个 profiling root 之间扫一遍 (sweep) 并对比。

Do not use this skill when

  • 任务是 HBM / 显存归因 → 用 ascend-memory-profiling
  • 任务是启停服务 → 用 vllm-ascend-serving
  • 任务是吞吐/性能 benchmark → 用 vllm-ascend-benchmark
  • 任务是采集新的 torch profiler 数据(起服务、控 profile 窗口、跑 workload、analyse) → 用 ascend-profiling-collection
  • profiling root 还没采到 kernel_details.csv(采集阶段失败) → 先回到 collection skill 排查;本 skill 不做补救。

Critical rules

  • 准确性优先于覆盖率:宁可报错或留 low confidence,也不输出无法追溯的结论。
  • 远端解析:profiling root 通常几十 GB,禁止全量拉回本地解析。本地只做静态检查、schema 校验、产物 manifest 阅读。真实 analyze 在远端容器里跑,必要时把 report/ 目录拉回本地。
  • 入口稳定:agent 调用 profile_analyze.py / profile_sweep.py,不要绕过去手写 python3 -m ascend_profile.analyze 命令。
  • manifest-aware:当 ascend-profiling-collection 产物可用时,优先把 --manifest <run_dir>/manifest.json 喂给 profile_analyze.py,让本 skill 自己从 manifest 里读 remote_profile_root / analysis_statusanalysis_status != "ok" 直接拒绝,不要静默跳过。
  • 进度协议:进度走 stderr,前缀 __VAWS_PROFILE_ANALYSIS_PROGRESS__=<json>。最终结果走 stdout,单个 JSON 对象。
  • 本地状态:本 skill 的本地状态全部放在 .vaws-local/profiling-analysis/runs/<timestamp>_<tag>/(untracked)。远端工作目录默认 /tmp/ascend_profile_framework
  • 不在算法里硬编码层数 / 模型语义:层数不能写成 Python 规则。已知模型的结构字段必须来自 config.json(显式提供、Hugging Face、ModelScope 或已登记本地 catalog)或已验证的 profile-visible hint;模糊族名(如 dsv4 / qwen3.5)必须枚举具体 variants 后逐个匹配,不能直接猜层数。量化和数据格式只影响权重大小、dtype 和效率解释,不作为层数/结构变体。

Read the full file on GitHub · 401 lines

Files

What ships with it

60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago Changed · +51 lines 0530cf2b982c
  2. 12d ago First seen · 350 lines · 191 tokens per session scan A db6ccd5064d2

Subscribe to this mod's changes

ascend-profiling-analysis is a skill published in the GitHub repository maoxx241/vllm-ascend-workspace (36 stars, last pushed 8d ago), licensed MIT. It adds 191 tokens to every session and 9,111 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…

microsoft/ai-agents-for-beginners · 200 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

insight-error-page

Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…

vercel/next.js · 83 tokens