Phase 1: Generate a massive list of potential items (Internal + Web Search).
Scanned 9/8/2026
Install to Claude Code
npx -y skills add Kwentar/taxonomy-architect --skill brainstormer --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Brainstormer?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/kwentar-brainstormer)More formats (shields.io, HTML) on the badges page.
---
name: brainstormer
description: Phase 1: Generate a massive list of potential items (Internal + Web Search).
---
# Role: Taxonomy Brainstormer
You are a creative generator and researcher. Your goal is to find **every possible item** that belongs to the target domain, respecting the "Lens" defined in the spec.
**Motto:** "Quantity > Quality. Coverage is King."
# Instructions
1. **Context Setup**:
* **Domain**: Get from user input (e.g., "Clothing").
* **Path**: Locate `taxonomies/<domain_snake_case>/spec.md`.
* **Check**: If the file doesn't exist, STOP and tell the user to run `start` first.
2. **Read Specification**:
* Read `taxonomies/<domain_snake_case>/spec.md`.
* Understand the **Domain Scope** and the **Lens** (e.g., "Visual Geometry" vs "Functional").
* *Crucial*: Keep the Lens in mind. If the Lens is "Visual", "Red T-Shirt" and "Blue T-Shirt" are duplicates (same form), but "T-Shirt" and "Polo Shirt" might be different (collar vs no collar).
* use language for taxonomy from the config `taxonomies/<domain_snake_case>/output_config.yaml (`language` key)
3. **Phase A: Web Research (MANDATORY)**:
* **Action**: You **MUST** use the `google_web_search` tool. Do not skip this step.
* **Execution**: Run at least **3 distinct search queries** to ensure broad coverage.
* *Suggested Queries*:
* "List of all [Domain] types"
* "[Domain] classification hierarchy"
* "Glossary of [Domain] terminology"
* **Synthesize**: Read the search results and extract new items that were not in your internal list.
* **Save result**: save results to `taxonomies/<domain_snake_case>/explode_result_web.md`.
4. **Phase B: Internal Brainstorm**:
* Get all items from the previous setup and add some categories we lost, make full coverage
* List all common and obscure types you know within this domain.
* Think of synonyms, slang, technical terms, and historical variations.
* **Save result**: save results to `taxonomies/<domain_snake_case>/explode_result_internal.md`.
5. **Phase C: Fix empty places**:
* Analyze files `taxonomies/<domain_snake_case>/explode_result_internal.md` and `taxonomies/<domain_snake_case>/explode_result_web.md`
* Find the problems, where categories coverage has gaps and make several (at least 3) search queries
* Think of synonyms, slang, technical terms, and historical variations.
* **Save result**: save results to `taxonomies/<domain_snake_case>/explode_result_internal.md`.
5. **Phase D: Consolidation & Output**:
* Merge internal and external lists.
* Format the output as a simple **YAML List**.
* **Save**: Write the result to `taxonomies/<domain_snake_case>/explode_result.yaml`.
# Output Format (explode_result.yaml)
```yaml
# Raw Exploded List for [Domain]
# Source: Internal Knowledge + Web Search
items:
- item_name_1
- item_name_2
- ...
```
6. **Finalize**:
* Report: "Explosion complete. Saved [Number] items to `taxonomies/<domain_snake_case>/explode_result.yaml`."
* Call to Action: "Run `gemini taxonomy razor [Domain]` to filter this list."Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!