Spreadsheet Ops (rsc-harness)

This is one skill from ericrisco's rsc-harness repo, a meta-tool that gives an AI coding agent memory, tools, skills and rules, with 269 skills in its catalogue; this entry covers spreadsheet-ops only. The skill opens with "a spreadsheet is a runtime, not a document": the agent's job is to hand over something checkable, an .xlsx that opens, a script that compiles, a cell range that can be read back, never a formula hand-waved in prose. It has the agent pick a tool by situation first: openpyxl to read and modify an existing local .xlsx, xlsxwriter for a new file rich in charts and conditional formats, pandas to dump a dataframe, gspread to read and write a Google Sheet from a script, the Sheets API batchUpdate for high-volume writes, and Apps Script for in-sheet menus and triggers. What sets it apart is that the common traps come first, for example that openpyxl stores formulas but doesn't compute them, so a cell reads back empty until Excel has opened and saved the file, along with Sheets API quotas and service-account permissions.

Who it's for:People who often hand Excel or Google Sheets work to an AI agent and have been burned by formulas and scripts before. Not useful if you don't code.

Before you start: Install an AI coding tool that supports Agent Skills, then add this one skill, or install the whole rsc-harness setup. The Python libraries it leans on (openpyxl, xlsxwriter, pandas, gspread) and any Google service account are yours to install and configure. Docs are in English. rsc-harness as a whole writes documentation, hooks and skills into your project; the README says it shows the plan and waits for your approval before writing anything, so read up on what it changes before a full install. The code is free under the MIT license.

Based on the project's public documentation. Not tested by AI Inspiration World.