cy-excel-mcp
Parses WeChat order messages, merges follow-up updates, and writes structured orders into a OneDrive Excel workbook through Microsoft Graph.
README
cy-excel-mcp
MCP server for parsing WeChat order messages, merging follow-up updates, and writing structured orders into a OneDrive Excel workbook.
Field mapping: docs/FIELD_MAPPING.md OpenClaw integration: docs/OPENCLAW_INTEGRATION.md Project roadmap: TODO.md
Overview
cy-excel-mcp is built for order-entry workflows where salespeople send order details through chat, sometimes as plain text, sometimes as follow-up updates after an image or partial draft.
The server converts those messages into structured JSON, merges updates into the same order, and writes the final result into a OneDrive Excel table through Microsoft Graph.
Typical Use Cases
- A salesperson sends a full text order and it should be recorded directly
- A salesperson sends part of an order first, then later sends the address, phone number, or payment info
- An OpenClaw Agent needs a stable MCP endpoint instead of relying on free-form chat reasoning
Example Input
单号:26.3.13-7
测试客户A
示例产品2000支105元
收件人: 测试联系人
手机号码: 13800000000
收货地址:测试省测试市测试区示例路88号
Example Output
{
"日期": "2026-03-13",
"单号": "26.3.13-7",
"销售员": "业务员A",
"客户": "测试客户A",
"货品名称": "示例产品",
"数量": "2000",
"数量单位": "支",
"销售金额": "105",
"总货款": "105",
"已收": "105",
"未收": "0",
"收货联系人": "测试联系人",
"收货人电话": "13800000000",
"收货地址": "测试省测试市测试区示例路88号"
}
Flow
- OpenClaw sends chat text into
ingest_order_message - The server parses the text into a normalized order object
- If this is a follow-up message, it merges the update into the existing draft
- The final order is matched against recent Excel rows
- The server updates an existing row or creates a new one
Privacy Note
This repository should only contain desensitized sample text and JSON. Do not commit real customer names, phone numbers, addresses, order screenshots, or payment details.
Features
- Parse order text from chat messages
- Merge follow-up text into an existing draft order
- Match recent orders by salesperson and customer from bottom to top
- Standardize product names against a OneDrive product catalog with local caching
- Write or update OneDrive Excel rows through Microsoft Graph
- Expose tools over Streamable HTTP for OpenClaw
Tools
ingest_order_messageparse_wechat_order_messagemerge_order_updateprocess_excel_ordercheck_product_catalog_statusrefresh_product_catalogresolve_product_nameanalyze_product_catalog_patterns
Requirements
- Python 3.10+
- A Microsoft Azure app registration with OneDrive permission
- An Excel workbook stored in OneDrive with a named table
Quick Start
git clone <your-repo-url> cy-excel-mcp
cd cy-excel-mcp
./bootstrap.sh
cp .env.example .env
Edit .env:
OC_OD_TENANT_ID=consumers
OC_OD_CLIENT_ID=your_microsoft_app_client_id
OC_OD_FILE_PATH=YourFolder/订单汇总.xlsx
OC_OD_TABLE_NAME=表1
OC_OD_CACHE_FILE=onedrive_token_cache.bin
OC_OD_PRODUCT_FILE_PATH=众一/2026诚亿报表.xlsx
OC_OD_PRODUCT_SHEET_NAME=产品明细
OC_OD_PRODUCT_CODE_COLUMN=A
OC_OD_PRODUCT_NAME_COLUMN=B
OC_OD_PRODUCT_CATEGORY_COLUMN=C
CY_PRODUCT_CACHE_FILE=product_catalog_cache.json
CY_PRODUCT_ALIAS_FILE=product_aliases.json
CY_EXCEL_MCP_HOST=127.0.0.1
CY_EXCEL_MCP_PORT=18061
CY_EXCEL_MCP_TRANSPORT=streamable-http
Start the server:
./start_cy_excel_mcp_http.sh
The default MCP endpoint is:
http://127.0.0.1:18061/mcp
Manual Installation
python3 -m venv .venv
. .venv/bin/activate
pip install --upgrade pip
pip install -e .
Or run directly after installation:
cy-excel-mcp --transport streamable-http --host 127.0.0.1 --port 18061
OpenClaw MCP Config
Example mcporter.json:
{
"mcpServers": {
"cy-excel-mcp": {
"baseUrl": "http://127.0.0.1:18061/mcp"
}
},
"imports": []
}
An example file is also included at config/mcporter.json.example.
Detailed runtime notes are available in docs/RUNNING.md.
Chinese runtime notes are available in docs/RUNNING.zh-CN.md.
OpenClaw integration notes are available in docs/OPENCLAW_INTEGRATION.md.
OpenClaw Agent Flow
- Full text order:
ingest_order_message - Follow-up text for an existing draft:
ingest_order_messagewithexisting_order - Parse only:
parse_wechat_order_message - Merge only:
merge_order_update - Write only:
process_excel_order
Why This Project
- Keeps Excel write logic outside prompt-only workflows
- Makes OpenClaw behavior more stable through explicit MCP tools
- Supports salesperson-first matching and bottom-up order updates
- Works well for chat-driven order entry teams using OneDrive Excel
Matching Rules
- Prefer matching within the same salesperson
- Search from the bottom of Excel upwards
- Match priority:
单号客户匹配客户别名
Repository Layout
cy_excel_mcp.py: MCP server implementationstart_cy_excel_mcp_http.sh: local HTTP startup scriptbootstrap.sh: one-command local setup.env.example: environment variable templateconfig/mcporter.json.example: OpenClaw MCP config template
Notes
- Personal OneDrive has been validated with
OC_OD_TENANT_ID=consumers. OC_OD_FILE_PATHcan include a nested folder path such asYourFolder/订单汇总.xlsx.OC_OD_TABLE_NAMEmust be the Excel table object name, not the worksheet name.- Product name standardization reads the product-code, product-name, and category columns from
OC_OD_PRODUCT_SHEET_NAMEinOC_OD_PRODUCT_FILE_PATH. - When a product name is matched, the matched product code is written to the order row field
编号if that column exists in the target order table. - Cup products use segmented matching for diameter, mold/series,
oz,ml, color, and material before generic text similarity. - Before each order write, the server checks product workbook metadata and uses
product_catalog_cache.jsonwhen the file has not changed. - The first Microsoft login may require device-flow authorization in the terminal.
- After the first successful sign-in, the token is cached in
onedrive_token_cache.binand reused on later runs. - Empty values do not overwrite existing Excel values.
- Keep
.envandonedrive_token_cache.binout of version control. - Temporary files such as swap files and
__pycache__are already ignored by.gitignore.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.