Agent Analytics

This Skill addresses the problems of fragmented product data analysis, cumbersome operations, and the difficulty in quickly identifying growth opportunities. In real-world development, product and growth teams frequently need to query data such as user behavior funnels, retention curves, path analysis, and experiment results. However, traditional tools either rely on SQL queries or require manually clicking through multiple UI layers, which is inefficient and prone to missing key insights.

Usage is straightforward. You simply need to run npx --yes @agent-analytics/cli@0.5.34 login in an AI Agent-supported editor (like Claude Code, Cursor) or terminal to complete browser authorization. After that, you can use natural language instructions to have the AI agent perform analysis tasks directly. For example, say "Analyze the user activation funnel for the past 7 days" or "Check the new user Day 1 retention trend." The AI will automatically invoke the corresponding CLI commands (e.g., funnel, retention) and return results. You can further ask follow-up questions like "Which step has the highest drop-off?" to receive growth suggestions.

It's well-suited for teams or individuals needing rapid product insights, especially product managers, data analysts, and growth operators who already use AI coding assistants in their daily development. Whether it's e-commerce, SaaS, or content platforms, as long as there's a need to track user behavior, optimize conversion paths, or evaluate experiment performance, this Skill can significantly shorten the time from question to decision.

It's recommended to use CLI commands uniformly for analysis instead of manually constructing API requests, as the CLI encapsulates all standard commands (e.g., funnel, retention, experiments) and automatically handles authentication and pagination. Note that the free tier has limits on data read per query. When encountering a PRO_REQUIRED error, you can run the upgrade-link command to guide users to upgrade to the Pro version. Additionally, this tool relies on a browser for login authorization; in headless runtimes (like Paperclip, OpenClaw), the --detached mode is required.