standard-workflow

Automate the standard operator generation workflow from design to validation.

258|48|Updated Jun 22, 2020
One-click install
npx skills add https://github.com/mindspore-ai/akg --skill standard-workflow
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: standard-workflow
Source: https://github.com/mindspore-ai/akg/tree/main/akg_agents/examples/run_skill/skills/standard-workflow
Command: npx skills add https://github.com/mindspore-ai/akg --skill standard-workflow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The standard operator generation workflow streamlines AKG Agent development by consolidating design, coding, and validation into a repeatable process, reducing iteration time and human error.

Core Features & Use Cases

  • Unified workflow for typical operators (e.g., matmul, conv2d) across common backends.
  • Clear coordination between coder-agent and verifier-agent to produce correct kernels and verify results.
  • Extensible structure that can accommodate additional child skills and verification steps as needs evolve.

Quick Start

Initialize the standard workflow by launching coder-agent to generate code and verifier-agent to validate the results.

Frequently Asked Questions about standard-workflow

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

FAQPage Schema
How do I automate operator generation and verification for AKG backends?

Automate operator generation for AKG backends by using a standard workflow that coordinates coder-agent code creation and verifier-agent validation, producing correct kernels while reducing iteration time.

What is the standard workflow for designing and validating AI operators?

The standard workflow for designing and validating operators consolidates design, coding, and verification into a repeatable process, utilizing coder-agent and verifier-agent to ensure correct kernel generation across common backends.

How do I coordinate coder-agent and verifier-agent to generate correct kernels?

Coordinate coder-agent and verifier-agent by initializing the standard workflow, where coder-agent generates the operator code and verifier-agent validates the results to ensure correct kernels for common backends.

Can I use this standard workflow for typical operators like matmul and conv2d across different backends?

Yes, you can use this standard workflow for typical operators like matmul and conv2d, as it provides a unified automation process for generating and validating these operators across common backends.

How do I extend the operator generation workflow with additional child skills?

Extend the operator generation workflow by adding additional child skills and verification steps into its extensible YAML structure, which requires defining the name and description frontmatter fields for each new child skill.

Why does my AKG agent workflow produce incorrect kernels during multi-stage generation?

Incorrect kernels often occur when the workflow lacks proper validation stages; utilizing a standard workflow with coordinated coder-agent and verifier-agent steps reduces human error and verifies results automatically.