skills.rest
mindspore-ai
mindspore-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

coder-agent

Generate executable code from algorithm designs for CUDA, Triton, and OpenCL.

258|48|Updated Jun 22, 2020
Search Tags:#optimization#ai-agent#code-generation#cuda#dsl#triton#opencl
One-click install
npx skills add https://github.com/mindspore-ai/akg --skill coder-agent-mindspore-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: coder-agent
Source: https://github.com/mindspore-ai/akg/tree/main/akg_agents/examples/run_skill/skills/coder-agent
Command: npx skills add https://github.com/mindspore-ai/akg --skill coder-agent-mindspore-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coder Agent converts algorithm designs into ready-to-run software, accelerating code generation and reducing manual implementation.

Core Features & Use Cases

  • Multi-DSL code generation: supports CUDA, Triton, and OpenCL to target diverse hardware.
  • Backend-aware optimization: suggests and applies backend-specific optimizations for NVIDIA, AMD, and Intel GPUs and CPUs.
  • Collaborative workflow: integrates with verifier-agent to validate correctness and quality.

Quick Start

Provide a design specification and specify the target DSL and backend to generate initial code.

Frequently Asked Questions about coder-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate CUDA or Triton code from an algorithm design?▼

To generate CUDA or Triton code, provide a design specification and specify the target DSL and backend. The system converts algorithm designs into ready-to-run software for prototyping and production deployment.

Can I use this to generate OpenCL code for AMD or Intel hardware?▼

Yes, you can generate OpenCL code for AMD and Intel hardware. Backend-aware optimization applies specific performance enhancements for NVIDIA, AMD, and Intel GPUs and CPUs.

What is the best way to accelerate software development across multiple GPU backends?▼

Accelerating software development across GPU backends is achieved by generating executable code from algorithm designs. This supports configurable workflows, reusable templates, and CI-friendly validation.

Does this code generation workflow support CI-friendly validation?▼

Yes, the code generation workflow supports CI-friendly validation. It integrates with verifier-agent to validate correctness and quality assurance for both prototyping and production deployment.

How do I verify the correctness of generated multi-DSL code?▼

To verify the correctness of generated multi-DSL code, use the collaborative workflow that integrates with verifier-agent. This validates correctness and quality assurance across specified backends.

What do I need to provide to start generating executable code from designs?▼

To start generating executable code, provide a design specification and specify the target DSL and backend. This generates initial code for your specific hardware environment.

Related Skills

Explore More Software Engineering Skills→
mindspore-aimindspore-ai
258

kernel-designer

Design UnifiedSketch algorithm sketches for operator tasks across multiple DSLs.

Official
Advanced
Supporter09Supporter09

llama-cpp

Deploys llama.cpp LLM inference on CPUs and non-NVIDIA hardware with GGUF quantization and OpenAI-compatible servers.

Community
Intermediate
BOHUYESHAN-APBBOHUYESHAN-APB
2

review-work

Coordinate a five-agent post-implementation review to validate code changes against goals and constraints.

Community
Advanced
kyo0221kyo0221

subagent-driven-development

Dispatch isolated subagents for multi-task implementation plans with mandatory two-stage reviews.

Community
Advanced