Draft practice-problem explanations from similar approved examples
At a glance
Adam Wise connected a Google Sheets question bank to Claude so new explanations could follow the style of his previously approved answers. This workflow is listed under education, retrieval-augmented-generation, spreadsheet-automation.
Agents / tools
Claude, Anthropic API, Google Sheets, Google Apps Script, Crammer API
His Google Apps Script reads a practice problem from Sheets and calls the Crammer API for similar problems. It combines the question and example explanations in a prompt, sends it to Anthropic's API and writes the generated answer back to the sheet. He describes a bank of more than 500 problems per course.
How it went
In the attached walkthrough, he describes getting more than 80% of the way to his desired explanation within seconds.
Worth knowing
The percentage is his assessment of draft completeness. He still reviews the explanations and adjusts wording before accepting them.
Save ideas in this browser. Upvotes count across the board.
The template rejects OTPs and unknown senders, structures transaction alerts into a Sheets ledger or review queue, and answers spending questions through a Claude Telegram agent.
The workflow reads language, proficiency, and interests from Sheets, researches current stories through Perplexity, and uses DeepSeek to produce level-adjusted reading, vocabulary, and grammar notes.
An indie hacker packaged an agency's eight repeated workflows (briefs, client approval with revision counters, quotes with margin checks, content calendars, post-mortems) as free MIT Claude skills.
Anthropic’s Building Effective Agents appendix highlights two proven patterns—support agents that pull data and take actions like refunds, and coding agents that iterate against tests—with clear success metrics and human review.