Prompt Library Architecture

A prompt library is a versioned collection of field-tested prompts organized by research domain and task type. Each prompt in the library has been used in production, evaluated against quality criteria, and refined based on output analysis. The library eliminates the cold-start problem — instead of engineering a new prompt for every research task, you select the closest template and customize it for the specific requirement.

The library has three organizational dimensions. Domain: competitive intelligence, technology evaluation, market sizing, regulatory analysis, customer research. Task type: collection, analysis, synthesis, verification, brief generation. Output format: executive summary, ranked assessment, comparison matrix, decision brief, threat profile. A prompt is indexed by all three dimensions. When a research task arrives, you select by domain and task type, then specify the output format. The customization is in the constraints and context — the structural architecture stays consistent.

## Prompt Library: Index Structure

DOMAIN: Competitive Intelligence
├── COLLECT: ci-landscape-scan-v3
├── COLLECT: ci-earnings-extraction-v2
├── COLLECT: ci-hiring-signals-v4
├── ANALYZE: ci-threat-assessment-v3
├── ANALYZE: ci-market-trajectory-v2
├── SYNTHESIZE: ci-quarterly-brief-v3
├── VERIFY: ci-cross-reference-v2
└── FORMAT: ci-board-presentation-v1

DOMAIN: Technology Evaluation
├── COLLECT: te-vendor-landscape-v2
├── ANALYZE: te-build-buy-partner-v3
├── ANALYZE: te-architecture-review-v2
├── SYNTHESIZE: te-decision-brief-v2
└── FORMAT: te-executive-summary-v1

Naming: [domain]-[task]-[version]
Each entry includes: prompt text, usage notes,
known limitations, quality criteria, example output.