Agent Skills by Ascend
Showing 77 vetted skills indexed across 1 GitHub repositories.
AscendvLLM-ascend_FAQ_Generator
Collect, classify, and generate Markdown FAQs from closed GitHub issues.
Ascendcatlass-operator-code-gen
Generate op_host, op_kernel, and test_aclnn scaffolds from CATLASS design documents.
Ascendascendc-operator-precision-eval
Generate and run AscendC operator precision tests, outputting Markdown and JSON reports.
Ascendascendc-operator-dev
Orchestrates seven-phase AscendC operator development from design to evaluation.
Ascendnpu-smi
Query and manage Huawei Ascend NPU status, health, and configuration via npu-smi.
Ascendascendc-mssanitizer
Analyze Ascend C operators for memory errors and generate reports.
Ascendtriton-operator-code-gen
Generate Triton kernels from Ascend NPU operator designs with smoke tests.
Ascendmegatron-commit-tracker
Normalize upstream Megatron-LM changes into deterministic change-set artifacts.
Ascendcatlass-operator-design
Translate CATLASS operator requirements into a structured design document template.
Ascendhccl-test
Automate HCCL performance testing across Ascend NPU clusters with MPI.
Ascendmodelscope-cli
Automates bulk ModelScope model and dataset downloads with SHA256 integrity checks and network diagnostics.
Ascendswanlab-setup
Automate SwanLab configuration, installation, and login across containerized environments.
Ascendmegatron-change-analyzer
Analyze Megatron-LM change-sets into structured feature evolution events.
Ascendsecurity-code-review
Detect common security vulnerabilities in Python, C++, Shell, and Markdown code.
Ascendverl-async-dapo
Generate Verl async DAPO training configurations with Megatron and FSDP2 options.
Ascendascendc-operator-code-gen
Generate AscendC operator host/kernel code from design docs and integrate into PyTorch.
Ascendascend-model-migration
Migrate and train open-source models on Ascend NPU with distributed training.
Ascendascendc-operator-code-review
Review Ascend C code snippets against explicit safety rules.
Ascendcatlass-operator-performance-optim
Optimize Catlass operators by adjusting tiling parameters and generating PRE/POST profiler reports.
Ascendtriton-operator-code-review
Analyze Triton operator code for static defects and Ascend NPU compliance.
Ascendskill-auditor
Audit AI agent skills and prompts for safety before deployment.
Ascendascendc-operator-testcase-gen
Generate validated test-case design plans for AscendC operators from design.md.
Ascendascend-docker
Provision Ascend NPU Docker environments with auto-mapped devices and volumes.
Ascendascendc-operator-doc-gen
Extracts AscendC operator interface data from source files into PyTorch-style README documentation drafts.
Frequently Asked Questions About Ascend
FAQPage SchemaWhat specific tasks can engineers perform using these Ascend skills?▼
Engineers can develop, profile, and optimize custom operators using AscendC or Triton, migrate models from GPU to NPU architectures, and manage distributed training configurations. The suite includes capabilities for precision debugging, memory error analysis, and automated documentation generation for operator interfaces.
Which technical personas are the primary users of these capabilities?▼
These skills are designed for AI infrastructure engineers, kernel developers, and machine learning researchers working on Huawei Ascend hardware. They are specifically tailored for professionals responsible for hardware-level performance tuning, model porting, and maintaining large-scale distributed training clusters.
What are the prerequisites for deploying these operator development skills?▼
Deployment requires a configured Linux environment with the CANN toolkit installed and accessible Ascend NPU hardware. Users must have the necessary driver and firmware versions verified via npu-smi, along with containerized environments for isolated operator compilation and testing.