nl2ledger

Natural language bookkeeping — parse Chinese/English expense descriptions and append to QianJi CSV. Trigger phrases: 记账, 记一笔, add expense, spent, 花了, 买了, lun...

This skill appears coherent and operates locally: it finds a QianJi_*.csv in your project root, shows a preview, and only writes after you confirm. Before installing or using it: 1) Ensure you have a backup of your ledger CSV (the script appends rows in-place). 2) Verify the CSV file the skill picks (pattern-based selection could pick the wrong file); if none is found, provide the correct path. 3) Pay attention to special rules in references/category_map.md (e.g., 桑拿/按摩 entries force 账户1=金色印象 and a default amount of 206.0) — these are policy choices that may create unexpected entries if you don't review the preview. 4) The script does not send data over the network and requires only python3; confirm python3 is available. 5) If you create a new CSV, ensure encoding/BOM expectations match your workflow (the schema mentions a UTF-8 BOM at the start). If you want stronger safety, manually inspect the preview before confirming each write or test the skill on a copy of your CSV first.

Type: OpenClaw Skill Name: nl2ledger Version: 1.0.0 The nl2ledger skill is a legitimate bookkeeping tool designed to parse natural language spending descriptions into a structured CSV format. The core logic in `scripts/append_entry.py` uses standard Python libraries to safely append data to a local file, and the `SKILL.md` instructions explicitly mandate a user confirmation step before any write operations occur.