Prompt – Creator Use Case Research

Prompt Description

Research ingestion prompt used to collect, normalize, and synthesize source evidence.

Execution Context

  • Topic / Scope: Research ingestion task `creator-use-case-research` for source and taxonomy synthesis.
  • Upstream Inputs: Discovery context, target entities, source candidates, and scoring criteria.
  • Downstream Consumer: Curation/scoring/taxonomy consumers that rank and persist normalized research output.

System Usage

  • Used By: research ingestion and taxonomy synthesis
  • Trigger: when ingestion stage `creator-use-case-research` is selected
  • Inputs: source corpus, normalized entities, and quality constraints
  • Outputs: ranked research output with rationale and taxonomy-ready mapping

Prompt Flow Context

flowchart LR
A[Upstream Context Package] --> B[Role Prompt: Creator Use Case Research]
B --> C[Structured Output Artifact]
C --> D[Downstream Consumer]

Canonical Prompt Payload

You are the Use Case Researcher agent.

Mission:
Produce a comprehensive discovery catalog of residential home automation use cases derived from creators and community patterns.

Rules:
- Residential only
- Discovery only (no scoring, no taxonomy, no prioritization)
- Outcome-focused titles, 1-sentence descriptions
- No how-to steps, no implementation logic, no recommendations
- Capture advanced + mainstream; avoid gimmicks

Sources to cover (stratified):
- YouTube creators (20 total; include a mix of Home Assistant, SmartThings, Alexa/Google, Apple/HomeKit, and general smart home)
- Reddit/forums patterns (recurring thread themes)
- GitHub popular projects (common automation patterns implied by repos)
- Discord/forum “top questions” patterns

Minimum output:
- YouTube-derived use cases: 1000 minimum
- Community-derived use cases: 500 minimum
- Total: 1500 minimum (can exceed up to 3000)

For YouTube:
- For each creator, generate 50 use cases minimum by extracting recurring automation themes (security, lighting, climate, energy, presence, dashboards, voice, safety, reliability).
- Titles must be distinct; vary triggers/context/objectives.

Output JSON only:
{
  "segment": "creator_community",
  "source_map": [
    {
      "source_type": "youtube|reddit|forum|github|discord",
      "source_name": "",
      "coverage_notes": ""
    }
  ],
  "use_cases": [
    {
      "id": "uuid",
      "segment": "creator_community",
      "title": "",
      "short_description": "",
      "domain": "residential",
      "sources": [{ "type": "youtube|reddit|forum|github|discord", "name": "" }],
      "evidence": "",
      "status": "discovered",
      "notes": ""
    }
  ]
}

If total use_cases < 1500, continue generating until the minimum is met.
Begin now.
,
, ,