product-decision-agent

Diagnose Chinese internet product problems and recommend prioritized next actions.

30.5k|3.5k|Updated Jul 4, 2025
One-click install
npx skills add https://github.com/davila7/claude-code-templates --skill product-decision-agent
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: product-decision-agent
Source: https://github.com/davila7/claude-code-templates/tree/main/cli-tool/components/skills/business-marketing/product-decision-agent
Command: npx skills add https://github.com/davila7/claude-code-templates --skill product-decision-agent

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Product managers and operators in Chinese internet companies face ambiguous, high-stakes questions—growth stalls, boss-inserted requirements, conflicting user feedback, project delays—and often get vague advice instead of a decisive judgment. This Skill acts like a senior product lead: it identifies the real blocker, the current stage, key constraints, and stakeholders, then gives the single most worthwhile next action.

Core Features & Use Cases

  • Structured Decision Output: Every answer follows a fixed format—problem judgment, cause analysis, 1-3 concrete actions with time windows and metrics, risk warnings, and at most 3 decision-changing questions.
  • 36 Scenario Playbooks: Covers requirement prioritization, roadmap planning, growth stagnation, CAC rise, retention drop, A/B test anomalies, cross-team conflict, OKR breakdown, postmortems, and more.
  • Reasoning Engine: A backstage workflow that separates facts from assumptions, locates the core bottleneck, judges the product stage, and defines conditions for scaling up, stopping, or rolling back.
  • Quality Gate Script: A Python checker that validates sample answers for Chinese-dominant language, concrete actions, decision signals, and absence of vague phrases.
  • Use Case: When your boss suddenly demands a loyalty points mall mid-sprint, the Skill tells you to present three trade-off options with cost and metric impact on one page, rather than silently absorbing scope.

Quick Start

Ask the agent in Chinese about a real product problem, such as "我们社区 DAU 连续两周下降 15%,运营想加活动拉回来,该怎么办?"

Frequently Asked Questions about product-decision-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prioritize product requirements when everything seems urgent?

Group requirements by the business result they change, not by who requested them. Identify which requirement removes the current core bottleneck, keep one main attack direction plus one evidence-gathering task, and compare conflicts by opportunity cost and reversibility rather than by who speaks loudest.

How should I handle a boss inserting a last-minute requirement?

Treat it as a resource conflict, not a simple feature request. Acknowledge the underlying goal, then present three options: keep the current version, run a 1-2 day validation, or swap out an equivalent scope. Write the delay, cut items, and metric impact on one page so the trade-off is explicit.

What should I check first when DAU drops suddenly?

First rule out data definition and tracking issues, then break the decline down by new versus returning users, channel, platform, version, and content supply. Act only on the largest contributing segment instead of launching a site-wide campaign that masks the real cause.

Can this Skill answer in English or only Chinese?

It answers in Chinese by default because it targets Chinese internet product work contexts, while keeping standard abbreviations like DAU, CAC, LTV, MVP, and OKR. English-led answers are explicitly discouraged in its instructions.

Why does the Skill limit clarifying questions to three?

The design requires giving a conditional judgment based on available information first, then asking only questions that would change the decision. This avoids dumping the problem back on the user while still flagging genuinely decision-critical unknowns.

What does the quality_gate.py script check in sample answers?

It validates that each candidate answer is Chinese-dominant, contains a problem judgment, concrete action hints, metric or decision signals, and risk guidance. It rejects vague phrases, banned source references, more than three confirmation questions, and files containing multiple aggregated answers.