Prompt Taxonomy Design

A prompt library without taxonomy is a junk drawer. Fifty prompts in a flat list means scrolling and guessing. A taxonomy organizes prompts into categories that match how the team works — by function (research, drafting, review), by deliverable type (proposal, brief, report), by department (sales, marketing, engineering), or by workflow stage (intake, analysis, output).

The taxonomy should have two levels maximum. Level one: category (e.g., "Sales Research"). Level two: specific prompt (e.g., "Competitive Positioning Analysis"). Three levels create navigation overhead. One level creates categories that are too broad. Two levels hit the sweet spot where a team member can find the right prompt in under 10 seconds.

1. 1. Inventory Existing Prompts Collect every prompt currently in use across the team. Ask each team member: "What are the 3-5 prompts you use most frequently?" Deduplicate and catalog the results.
2. 2. Identify Natural Categories Group the collected prompts by function or deliverable type. The categories should emerge from actual usage patterns, not from an abstract organizational theory. If 80% of prompts are research-related, "Research" gets subcategories.
3. 3. Define Metadata Standards Every prompt gets: name, category, description (one sentence), author, version, last tested date, and required context documents. Metadata makes prompts discoverable and auditable.