kunpengcompute
- 29 skills
- 0 followers
- 6 hours ago last updated
- ▌ Kernel Patch · kunpengcompute bundle内核补丁迁移与冲突解决工作流。用于导出 Git 提交为 patch、检查补丁是否已存在、分析 .rej 冲突、按顺序批量合入 patch_sets，并在每个补丁合入后执行确定性 post-apply 门禁校验与自动复检循环。适用于 Linux 内核补丁迁移、config 文件映射同步、顺序合入多个补丁集、补丁边界漂移检查和补丁合入质量校验。
- ▌ Patch Validator · kunpengcompute bundle校验 git format-patch 格式的补丁文件与目标 git 仓库的合入情况。支持两种模式：(1) 应用校验，逐个分析每个补丁的每个 hunk 是否可以应用到目标分支；(2) 已合入差异校验，将本地补丁与目标仓库中已合入提交的 diff 做逐行比较，识别缺失行、额外行和删除侧不一致，并生成详细的 Markdown 格式报告。当用户需要验证补丁是否可以应用到目标仓库、检查本地补丁与上游已合入补丁是否完全一致、或生成补丁状态报告时使用此技能。
- ▌ Run Dev · kunpengcomputeBuild, test, format, and develop CubeSandbox components. Use when asked to build, compile, run tests, format code, or do development work on this project.
- ▌ Sdk E2e Cases · kunpengcompute bundleWrite, review, and execute CubeSandbox SDK compatibility E2E pytest cases. Use when the user asks to add, design, review, debug, or run SDK E2E cases under tests/e2e/sdk_compat, or mentions lifecycle, network policy, sandbox templates, backend compatibility, pytest markers, or live E2E validation.
- ▌ Cube Sandbox · kunpengcompute bundleCube Sandbox 安全沙箱执行技能。当需要安全执行 Python 代码、Shell 命令，或需要隔离环境运行不受信任的代码时使用。 适用场景： (1) 用户要求"在沙箱中执行代码"、"跑一段 Python"、"安全执行"、"隔离环境运行"； (2) 需要执行可能有副作用的代码（文件操作、网络请求、安装包等），希望通过沙箱隔离风险； (3) 需要读写沙箱内文件、挂载宿主机目录到沙箱； (4) 需要控制沙箱网络策略（完全断网、白名单、黑名单）； (5) 需要暂停/恢复沙箱（保留内存快照以实现快速复用）。 Cube Sandbox 兼容 E2B SDK，通过环境变量 E2B_API_URL 指向 Cube 部署地址即可使用。
- ▌ Generate SQL Testcases · kunpengcompute bundle为Spark表达式生成测试SQL用例。读取包含Spark函数信息的CSV文件，为每个函数生成19种数据类型的测试SQL并输出到新CSV。使用场景：用户需要为Spark函数生成测试SQL、处理CSV中的表达式列表、批量生成SQL测试用例、或涉及OmniOperator/gluten-velox表达式兼容性测试。触发关键词包括但不限于：生成测试SQL、表达式测试、SQL测试用例、Spark函数测试、批量生成SQL、数据类型测试、Spark表达式CSV、OmniOperator测试、gluten-velox测试。即使用户没有明确提到"skill"或"脚本"，只要涉及从CSV中的函数列表生成SQL测试，就应使用此skill。
- ▌ Omnioperator Expression Dev · kunpengcompute bundleUse for OmniOperator expression development. Trigger when implementing new vectorized expressions, functions, or operators in OmniOperator framework. This includes scalar functions, aggregate functions, window functions, variadic functions, bitwise operations, math functions, string functions (like concat, split, substring), or any SQL function implementation. Also trigger when asked to "implement function", "add expression", "create vectorized function", "实现函数", "添加表达式", "编写函数", when working with Register*.cpp registration files, or when user mentions "OmniOperator" with "function/expression/implementation" or "函数/表达式/实现" and specifies function signature, arguments, return type, NULL handling, or vector encoding support.
- ▌
- ▌
- ▌ Ov Dream · kunpengcompute bundleUse when the user explicitly types `ov dream` or `ov recall <query>` and the request should be routed to the OpenViking sync/recall CLI instead of handled as normal chat.
- ▌
- ▌ Ov Add Data · kunpengcomputeThis skill adds data(like resources) to OpenViking Context Database (aka. ov). Use when an agent needs to add files, data from URLs, or external knowledge during interactions. Trigger this tool when 1. is explicitly requested adding files or knowledge; 2. identifies valuable resources worth importing; 3. the user mentioned adding to OV/OpenViking/Context Database. This skill helps how to use CLI like `ov add-resource`, `ov add-skill` and `ov add-memory` to add resource data, skill files, memory files to OpenViking.
- ▌
- ▌ Ov Search Context · kunpengcomputeSearch context data(memories, skills and resource) from OpenViking Context Database (aka. ov). Trigger this tool when 1. need information that might be stored as memories, skills or resources on OpenViking; 2. is explicitly requested searching files or knowledge; 3. sees `search context`, `search openviking`, `search ov` request.
- ▌ Openviking · kunpengcomputeActivate when the user asks about any repository listed in the system prompt under 'OpenViking — Indexed Code Repositories', or when they ask about an external library, framework, or project that may have been indexed. Also activate when the user wants to add, remove, or manage repos. Always search the local codebase first before using this skill.
- ▌ Install Openviking Memory · kunpengcomputeInstall and configure the OpenViking long-term memory plugin for OpenClaw via natural conversation. Once installed, the plugin automatically captures important facts from chats and recalls relevant context before each reply (auto-capture + auto-recall, cross-session). Covers prerequisites check, plugin install through OpenClaw's plugin manager first, with ov-install only as a backup path, wizard-based configuration, slot activation, gateway restart, verification, plus multi-tenant root-key support, multi-instance, and uninstall. Trigger when the user says any of: "install OpenViking", "set up memory", "configure memory plugin", "add long-term memory", "connect to OpenViking server", "RAG", "semantic memory", "帮我装 OpenViking", "配置记忆插件", "安装记忆功能", "接入 OpenViking", "我有一台 OpenViking 服务器". The user does NOT need to know any CLI commands — the agent runs everything and only asks for a few values. This skill assumes the OpenViking server is already running. If the server is not ready, the skill tells the user to con
- ▌ Add Jit Kernel · kunpengcomputeStep-by-step tutorial for adding a new lightweight JIT CUDA kernel to sglang's jit_kernel module
- ▌ Add Sgl Kernel · kunpengcomputeStep-by-step tutorial for adding a heavyweight AOT CUDA/C++ kernel to sgl-kernel (including tests & benchmarks)
- ▌ CI Workflow Guide · kunpengcomputeGuide to SGLang CI workflow orchestration — stage ordering, fast-fail, gating, partitioning, execution modes, and debugging CI failures. Use when modifying CI workflows, adding stages, debugging CI pipeline issues, or understanding how tests are dispatched and gated across stages.
- ▌ Clean Startup Log · kunpengcomputeClean up noisy startup warnings and spurious prints in SGLang server logs. Use when users ask to clean up unwanted warnings, deprecation messages, or third-party noise in the server startup output.
- ▌ Write Sglang Test · kunpengcomputeGuide for writing SGLang CI/UT tests. Covers CustomTestCase, CI registration, server fixtures, model selection, mock testing, and test placement. Always read test/README.md for the full CI layout, how to run tests, and extra tips. Use when creating new tests, adding CI test cases, writing unit tests, or when the user asks to add tests for SGLang features.
- ▌ Sglang Sota Performance · kunpengcomputeEnd-to-end SGLang SOTA performance workflow. Use when a user names an LLM model and wants SGLang to match or beat the best observed vLLM and TensorRT-LLM serving performance by searching each framework's best deployment command, benchmarking them fairly, profiling SGLang if it is slower, identifying kernel/overlap/fusion bottlenecks, patching SGLang code, and revalidating with real model runs.
- ▌ LLM Serving Auto Benchmark · kunpengcompute bundleFramework-independent LLM serving benchmark skill for comparing SGLang, vLLM, TensorRT-LLM, or another serving framework. Use when a user wants to find the best deployment command for one model across multiple serving frameworks under the same workload, GPU budget, and latency SLA.
- ▌ LLM Torch Profiler Analysis · kunpengcompute bundleUnified LLM torch-profiler triage skill for `sglang`, `vllm`, and `TensorRT-LLM`. Use it to inspect an existing `trace.json(.gz)` or profile directory, or to drive live profiling against a running server and return one three-table report with kernel, overlap-opportunity, and fuse-pattern tables.
- ▌ Sglang Diffusion Add Model · kunpengcompute bundleUse when adding a new diffusion model or Diffusers pipeline to SGLang.
- ▌ Sglang Diffusion Performance · kunpengcomputeUse when choosing the fastest SGLang Diffusion flags for a model, GPU, and VRAM budget.
- ▌ Sglang Diffusion Ako4all Kernel · kunpengcompute bundleUse when optimizing an existing SGLang diffusion kernel with AKO4ALL, including AKO4ALL repo hygiene, custom microbench setup, ncu-guided iteration, and end-to-end denoise validation. Also use when a sibling AKO4ALL repo must be cloned or refreshed before starting kernel tuning work.
- ▌ Sglang Diffusion Modelopt Quant · kunpengcomputeUse when quantizing a diffusion DiT with NVIDIA ModelOpt and making the resulting FP8 or NVFP4 checkpoint loadable, verifiable, and benchmarkable in SGLang Diffusion.
- ▌ Sglang Diffusion Benchmark Profile · kunpengcompute bundleUse when benchmarking denoise latency or profiling a diffusion bottleneck in SGLang.