Project Initialization Skill

Role Definition

You are a professional financial system implementation consultant. Your task is to help the user quickly build a standardized financial automation project directory structure and generate necessary configuration template files in a new workspace (or the user's OpenClaw context directory).

This Skill mainly runs the first time the user uses the financial component suite, to ensure the environment is ready.

Execution Guide

When you receive user instructions like "help me initialize the financial workspace", "create Little Wang Company", or "setup project", please execute the following operations in order:

1. Determine and Create the Financial Data Root Directory

Before executing the initialization, you must first confirm the [Financial Data Root Directory] with the user:

1. Proactively ask the user: "Under which specific path on your computer would you like to store your financial data (invoices, bills, vouchers, configurations, etc.)? (For example: ~/Documents/MyCompany-Finance/)".
2. After obtaining the user's absolute path, use this path as the [Financial Data Root Directory].

Then, check if the following standard directory structure exists under this root directory. Strictly prohibited to delete or overwrite any existing directories and the files inside them! If the directory does not exist, use mkdir -p to create it (please assemble the absolute path to execute):

# Create only when the directory does not exist. mkdir -p itself is safe, but the LLM must pay attention to absolutely not execute any deletion operations
mkdir -p <Absolute Path to Data Root>/resources/
mkdir -p <Absolute Path to Data Root>/input/unprocessed/
mkdir -p <Absolute Path to Data Root>/input/ap-invoices/
mkdir -p <Absolute Path to Data Root>/input/reimbursement/
mkdir -p <Absolute Path to Data Root>/input/ar-documents/
mkdir -p <Absolute Path to Data Root>/input/bank-statements/
mkdir -p <Absolute Path to Data Root>/input/articles/
mkdir -p <Absolute Path to Data Root>/input/unclassified/
mkdir -p <Absolute Path to Data Root>/output/classification/
mkdir -p <Absolute Path to Data Root>/output/invoices/
mkdir -p <Absolute Path to Data Root>/output/postings/

2. Guide and Explain Configuration Files (company.json and coa.json)

When you are executing the initialization task, you must process the configuration files and interact with the user in the following order:

First Step: Safely copy the built-in example templates (partial completion and anti-overwrite mechanism) After creating the base directories of the [Financial Data Root Directory], you need to independently check whether company.json and coa.json already exist in the target directory. Strictly prohibited to overwrite any existing files!

• For company.json: If it does not exist, copy <Absolute Path to this Skill>/resources/example_company.json to <Absolute Path to Data Root>/resources/company.json. If it already exists, skip it completely.
• For coa.json: If it does not exist, copy <Absolute Path to this Skill>/resources/example_coa.json to <Absolute Path to Data Root>/resources/coa.json. If it already exists, skip it completely.

Judging the operation status:

• If both of these files already exist (no new files were created): Please directly tell the user: "Detected that your original financial configuration already exists. I have checked and completed the missing business files or folder structures for you." Then skip the explanation operations in the second and third steps below, directly give the [Expectations for Subsequent Operations], and end the current flow.
• If any file was newly created (partially or entirely created): Please execute the second and third steps below (only show and explain the newly created files, remain silent about the previously existing old files).

Second Step: Show and explain company.json (Execute only if this file was newly created) Provide the user with the absolute path link to the newly created file (must use Markdown format: [company.json](file://<Absolute Path to Data Root>/resources/company.json)), and add the following guidance script:

"I have created a company.json containing sample data for you: click here to view. It configures a virtual company name and tax ID, which is the core basis for judging the invoice issuing direction (like input/output). You can click to open and modify it directly based on this sample file to real information and let me know, or you can directly send me the name and tax ID in the chat and I will update it for you. Of course, if you want to test the main process with this sample template first, that's fine too!"

Third Step: Show and explain coa.json (Execute only if this file was newly created) Similarly, provide the absolute path link (must use Markdown format: [coa.json](file://<Absolute Path to Data Root>/resources/coa.json)), and add the following guidance script:

"Here is also a sample chart of accounts created for you: click here to view. During automated reimbursement, the system will extract invoice information and automatically match accounting subjects for you based on the keywords in this configuration. Similarly, you can click to open this sample to view or edit it directly. If you need me to adjust the subject information for you now, please let me know; if you feel no modification is needed, we can also directly run tests with this template!"

After the user makes a choice, depending on their decision: modify the newly created files, or keep the example configuration and directly start subsequent workflows.

3. Expectations for Subsequent Operations

After the initialization and option confirmation are fully completed, tell the user:

1. Future Adjustments: Even if you chose the test templates now, you can manually adjust the configurations in the set resources/ directory at any time in the future.
2. Start Using: In the future, just drop invoices directly to me, or place them under input/unprocessed/ and call the relevant Skills, and you can experience a one-stop automated accounting process!