# Human Signal: research interface

Product domain: https://humansignal.studio

Interface proposal for commissioning qualitative and quantitative research through MCP. This document is a contract example, not a connection guide: no hosted endpoint, authentication flow, or ordering backend is implemented in this local website.

## Workflow

1. `request_quote`: provide the decision, material, target audience, method, budget limit, and requested deliverables. Return an itemized quote, scope, delivery terms, and expiry. No purchase occurs at this step.
2. `confirm_order`: reference the quote ID and authorize the agreed total. Require explicit approval from the party paying. An idempotency key prevents duplicate orders.
3. `get_status`: retrieve recruitment, fieldwork, or delivery status for an authorized order.
4. `get_results`: retrieve participant evidence or response data, findings, and the study methodology. Keep original evidence separate from interpretation.

## Research choices

- Qualitative: page reviews, interviews, and usability sessions. A page clarity review covers one page, five participants, and three questions. Larger studies are scoped individually.
- Quantitative: concept preference, message testing, and audience comparisons. Define recruitment, sample size, quotas, exclusions, questionnaire, and analysis before fieldwork.

## Example quote request

```json
{
  "method": "qualitative",
  "decision": "Can first-time visitors identify the product and next step?",
  "material": {"type": "landing_page", "page_count": 1},
  "audience": {"description": "Small-team project leads", "language": "en"},
  "scope": {"participants": 5, "questions": 3},
  "deliverables": ["participant_responses", "findings", "methodology"]
}
```

Budget limits use an explicit currency and amount. Prices and delivery times must come from the quote, not from examples in this document.

## Outputs

`sample-report.json` contains a qualitative fixture with questions, fictional participant responses, evidence references, and interpretation.

`sample-quantitative.json` contains a quantitative fixture with synthetic counts, denominators, percentages, and limitations. These are aggregate demonstration data, not a participant dataset or a completed study.

Both files explicitly mark the data as illustrative. No study was conducted.

## Implementation requirements

Before enabling live tools, define authentication and order-level access, input/output schemas, error codes, quote expiry, cancellation/refund rules, participant consent and privacy handling, and retry behavior. Do not place orders without the payer's approval. Do not expose participant contact details in research outputs.