Trade Pipeline Init

Trade Pipeline Init — Interactive Setup Skill

Trigger

When the user says "初始化配置", "setup", "configure trade pipeline", "init", "第一次使用", or when trade_pipeline/config/config.yaml contains example/placeholder data (seller name is "ACME EXPORT").

Behavior

Use AskUserQuestion to guide the user through configuring their trade pipeline. Collect all information first, then write config.yaml in one step.

Step 1: Company Info

Ask with AskUserQuestion (multiSelect: false):

• "你的公司英文名？" — free text input (user selects "Other")

Then ask:

• "公司中文名（可选）？"
• "公司地址？"
• "联系人姓名？"
• "联系邮箱？"
• "联系电话？"

These can be combined into 2-3 rounds of AskUserQuestion, grouping related fields.

Step 2: Trade Terms

Ask with AskUserQuestion options:

question: "默认贸易条件？"
options:
  - label: "FOB"
    description: "Free On Board — 卖方交到港口，买方负责运输"
  - label: "CIF" 
    description: "Cost, Insurance, Freight — 卖方负责运费+保险到目的港"
  - label: "DDP"
    description: "Delivered Duty Paid — 卖方全包到买方门口"
  - label: "EXW"
    description: "Ex Works — 买方自提"

Step 3: Currency & Port

Ask with AskUserQuestion options:

question: "默认币种？"
options:
  - label: "USD"
  - label: "CNY"
  - label: "EUR"

question: "默认装运港？"
options:
  - label: "QINGDAO,CHINA"
  - label: "SHANGHAI,CHINA"
  - label: "NINGBO,CHINA"

Step 4: Payment & Lead Time

Ask:

• "付款条件？" (default: "30% T/T deposit; 70% before shipment")
• "交货期？" (default: "45-60 days after deposit")

Step 5: First Buyer (Optional)

Ask:

question: "现在添加第一个客户吗？"
options:
  - label: "是，添加客户"
  - label: "跳过，稍后再加"

If yes, ask buyer name, address, contact, email.

Step 6: Write Config

After collecting all info, generate the YAML config and write to: trade_pipeline/config/config.yaml

Use the following structure:

sellers:
  :
    name_cn: 
    name_en: 
    address: 
    contact: 
    tel: 
    email: 
    bank: { name: "", swift: "", account_no: "" }

buyers:
  :  # if provided
    name_en: 
    legal_names: []
    aliases: []
    address: 
    contact: 
    email: 

format_defaults:
  standard:
    seller_id: 
    currency: 
    price_unit: "/PC"
    terms_id: "default_"

terms_templates:
  default_:
    payment: 
    delivery: " , Incoterms 2020."
    lead_time: 
    validity: "10 days"
    packing: "Standard export packaging: 25 kg cartons, Euro pallets."
    quality: "100% inspection before shipment."

defaults:
  port_of_loading: 
  pi_number_pattern: "PI-{order_no}"
  ci_number_pattern: "CI-{order_no}"
  quote_no_pattern: "QT-{order_no}"
  date_format: "%d %B %Y"

packing:
  carton_weight_kg: 25
  pallet_self_weight_kg: 28
  cartons_per_pallet: 36

pl_profiles:
  default:
    pl_config: standard
    packing_profile: standard_25kg

cache:
  dir: .cache/understanding
  prompt_version: v1.0
  schema_version: v1.0
  enabled: true

ocr_review:
  force_review_fields: [description, standard, quantity, unit, weight_kg]
  confidence_threshold: 0.90

Seller ID generation: take the first 2 meaningful words from the English name, lowercased, joined by underscore. Skip words like "co", "ltd", "inc", "limited".

Step 7: Confirm

Show the user a summary:

Seller:  (id: )
Terms:   | Currency:  | Port: 
Buyer:   (if added)

Config written to: trade_pipeline/config/config.yaml

Quick demo:
  python -m trade_pipeline --input examples/sample_inquiry.xlsx --order DEMO --buyer 

Notes

• The CLI version (python -m trade_pipeline init) is still available as a fallback
• This skill provides a better UX within Claude Code using native dialog controls
• Bank info can be added later by editing config.yaml directly

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Dangooy
• Source: Dangooy/trade-pipeline-skill
• License: MIT

Install and usage instructions live in the source repository linked above.