AI Services Market › Use cases › Work & Operations
Work & OperationsDone with Claude Code 2026-10-20
Generate recurring weekly reports combining KPIs and pull requests
A product leader uses Claude Code for automatic recurring reports on what developers shipped and whether features helped or hurt KPIs.
- What They Did
- David Kelly, a product leader, tracks KPIs for each product in a Metabase dashboard. He prompts Claude Code to build a system that pulls weekly GitHub pull requests, compares them with period-over-period KPI changes, and scores how likely each PR affected each KPI. He connects Zen MCP tools to the OpenAI and Gemini APIs so several models cross-check the analysis. He publishes the full prompt.
- How It Went
- He gets a weekly report on whether shipped features helped or hurt KPIs, and he uses it to iterate. The post gives no accuracy figures for the correlations.
- Worth Knowing
- A paid Claude plan is needed. He suggests Pro at $20 a month and advises against the API because costs can add up quickly.
Try it yourself with Claude Code
Every [Monday morning], pull the pull requests merged in [repo name] over the past week and our KPI numbers from [KPI source, e.g. a CSV or dashboard export]. Write a one-page report of what shipped, which features likely helped or hurt [KPI names], and what to watch next week. Save it as [file path or format] and show me the first draft before setting up any recurring schedule or sending it to anyone.
Do this with a skill.
These AI Services Market playbooks do this exact job.