For analysis projects: studies, literature reviews, hypothesis tracking.
Choose this preset when your project:
- Has
concepts/,hypotheses/,bibliography/, oranalysis/ - Tracks research questions or hypotheses
- Requires rigorous evidence standards
## Research Quality Standards
- Evidence tier classification for all claims (Tier A-D)
- Hypothesis tracking with confidence levels
- Source attribution and citation
- Document contradictions and limitations
- Reproducible methodology| Component | Tier | Recommended | Why |
|---|---|---|---|
| permissions.allow | 1 | ✅ Yes (baseline) | Pre-approve git read commands |
| Stop hook | 1 | ✅ Yes (baseline) | Uncommitted/unpushed reminders |
| CLAUDE.md | 2 | ✅ Yes | Project context, evidence tiers |
| Session hook | 2 | ✅ Yes | Shows research status |
| Post-tool hook | 2 | Optional | Auto-update indexes if tracking |
| GitHub Actions | 3 | For teams | @.claude research reviews |
See Project Infrastructure Pattern for the full tiered approach.
Include in CLAUDE.md:
## Evidence Tiers
- **Tier A**: Primary sources, production data, official specs
- **Tier B**: Peer-reviewed, expert interviews, validated analysis
- **Tier C**: Industry reports, vendor docs, practitioner blogs
- **Tier D**: Opinions, speculation (label as such)
Strong claims require Tier A or B evidence.Consider adding:
## Active Hypotheses
Track research hypotheses with:
- Clear, falsifiable statement
- Confidence level (1-5)
- Evidence tier of supporting data
- Validation method## Git Workflow
Commit messages:
- `research:` or `📊` New findings or analysis
- `docs:` or `📚` Documentation updates
- `data:` Data processing changes
- `hypothesis:` Hypothesis updates
- `cite:` Bibliography additionsSee research project examples in the patterns documentation.