research-architect-designlisted
Install: claude install-skill mmTheBest/Research-Architect
# First Paper Study Design
Use this branch after `confirmed_research_spine.md` exists.
## Blocking rule
If the confirmed research spine is missing, return to `research-architect-literature` and obtain user confirmation before designing the study.
## Inputs
Read:
```text
paper_output/confirmed_research_spine.md
paper_output/sota_gap_map.md
paper_output/source_inventory.md
paper_output/project_config.json
```
## Required outputs
Create:
```text
paper_output/study_design_brief.md
paper_output/analysis_plan.md
paper_output/evaluation_plan.md
paper_output/experiment_registry.md
paper_output/threats_to_validity.md
```
## Study design table
| Component | Decision | Rationale | Claim Supported | Risk | Mitigation |
|---|---|---|---|---|---|
| Research question | | | | | |
| Unit of analysis | | | | | |
| Data/material | | | | | |
| Method/approach | | | | | |
| Baseline/control | | | | | |
| Evaluation criterion | | | | | |
| Robustness check | | | | | |
| Failure/negative case | | | | | |
| Interpretation boundary | | | | | |
## Evidence modes
Choose the design logic that matches the project:
- computational;
- experimental;
- theoretical;
- qualitative;
- design/build/evaluation;
- mixed-methods.
Do not force all projects into the same template.
## Rules
- Every experiment, simulation, analysis, proof, or evaluation must support a specific claim or scope boundary.
- Every central claim needs at least one direct evidence