What problem does it solve? Working with Feishu (Lark) spreadsheets programmatically is error-prone: formulas behave differently from Excel, writes can silently lose data types, and edits are hard to verify. This Skill gives an AI agent a complete, rule-driven workflow for reading, writing, formatting, and validating online spreadsheets through the lark-cli command line. ## Core Features & Use Cases - Full spreadsheet operations: Create workbooks, manage sheets and row/column structure, read and write cell values, formulas, styles, comments, and images, plus search/replace and atomic batch updates. - Data objects and visualization: Build pivot tables, charts, conditional formats, filters, filter views, sparklines, and floating images as real native objects that stay linked to source data. - Formula correctness pipeline: Translate Excel formulas into Feishu-compatible semantics (ARRAYFORMULA, MAP/LAMBDA, date differences) and verify zero-error results with the formula-verify self-check. - Use Case: Ask the agent to build a financial model in a Feishu sheet—it will create the workbook with typed data, apply styling, write verified formulas, add a pivot summary and chart, then read back the results to confirm everything took effect. ## Quick Start Ask the agent to read the Feishu spreadsheet at a given /sheets/ URL and summarize its sheets, then write or update data in a specified range.