AgentSkillsCN

sku_analyser

阶段 1：将对象关键词解析为具体的资产 ID，并获取其尺寸数据。适用场景：(1) 用户仅以常见名称提及对象，而未提供具体 ID；(2) 需要将自然语言中的对象描述映射为网格或资产标识符；(3) 为空间规划获取对象尺寸信息。系统会搜索资产数据库，返回精确的标识符。

SKILL.md
--- frontmatter
name: sku_analyser
description: "Stage 1: Resolves object keywords to specific asset IDs and retrieves dimensional data. Use when: (1) User mentions objects by common names without specific IDs, (2) Need to map natural language object descriptions to mesh/asset identifiers, (3) Require object dimensions for spatial planning. Searches asset database and returns precise identifiers."
metadata:
  stage: "1"

SKU Analyzer

Overview

Translates common object names into specific asset IDs with dimensional metadata. Bridges natural language object references with technical asset identifiers.

Usage

bash
python scripts/analyze_sku.py --keyword "apple"

Supported Categories

  • •Fruits: apple, orange, banana
  • •Containers: box, bin (small/medium/large)
  • •Bottles: bottle, water bottle
  • •Shapes: cube, cylinder, sphere
  • •Tools: screwdriver, wrench
  • •Furniture: table, workbench, shelf
  • •Machinery: conveyor

Example Mappings

  • •"apple" → mesh_fruit_apple_01 [0.08, 0.08, 0.09] 0.2kg
  • •"box" → container_cardboard_medium [0.3, 0.3, 0.2] 0.5kg
  • •"bottle" → bottle_plastic_500ml [0.07, 0.07, 0.2] 0.5kg

Input Schema

{ "object_keyword": "string" }

Output Schema

{ "sku_id": "string", "dimensions": [0.1, 0.1, 0.1] }