Files
kesslerio_academic-deep-res…/README.md

121 lines
3.8 KiB
Markdown

# Academic Deep Research 🔬
**Transparent, rigorous, self-contained research** — not a black-box API wrapper.
## Why This Skill Exists
Most "deep research" tools are wrappers around external APIs. You send a query, get a report, and have no idea what happened in between.
**This skill is different:**
-**Full methodology visible** — Every step documented, reproducible
-**No external dependencies** — Runs entirely on OpenClaw native tools
-**User control** — 3 explicit checkpoints for approval
-**Academic rigor** — APA citations, evidence hierarchy, confidence levels
-**Works offline** — No API keys, no cloud services
## Comparison with Cloud-Based Research Tools
| Feature | This Skill | Cloud API Wrappers |
|---------|------------|-------------------|
| Methodology | Fully documented | Black box |
| Dependencies | None | External API + key |
| Offline | ✅ Yes | ❌ No |
| User Checkpoints | 3 approval points | Usually none |
| Citation Format | APA 7th edition | Varies/unspecified |
| Evidence Hierarchy | Explicit (meta-analyses → opinion) | Unspecified |
| Output Control | Strict prose, no bullet points | Varies |
| Reproducibility | ✅ Same inputs = same process | ❓ Unknown |
## Core Features
### Mandated Research Cycles
Every theme gets **minimum 2 full research cycles**:
1. Broad landscape search → Analysis → Gap identification
2. Targeted deep dive → Challenge assumptions → Synthesis
No shortcuts. No single-pass summaries.
### Evidence Standards
- **Every conclusion cites multiple sources**
- **Contradictions must be addressed** — not hidden
- **Confidence annotations:** [HIGH], [MEDIUM], [LOW], [SPECULATIVE]
- **Evidence hierarchy:** Meta-analyses > RCTs > Observational > Expert opinion
### Academic Output
- Flowing narrative prose (no bullet point dumps)
- APA 7th edition citations (1-2 per paragraph)
- Proper paragraph structure: claim → evidence → analysis → transition
- Executive summary, methodology, findings, limitations, references
### User Control
Three mandatory stop points:
1. **Initial Engagement** — Clarify scope before research
2. **Research Planning** — Approve themes and approach
3. **Final Report** — Review completed analysis
## Quick Start
```
/research "Comprehensive analysis of [your topic]"
```
Or just ask for "deep research on..." or "exhaustive analysis of..."
## Research Protocol
### Phase 1: Clarification
Agent asks 2-3 essential questions, confirms understanding, **waits for you**.
### Phase 2: Planning
Agent presents:
- Major themes identified (3-5)
- Research execution plan (table format)
- Expected deliverables
**You approve before execution begins.**
### Phase 3: Execution (Auto)
For each theme, two full cycles:
- `web_search` (count=20) for landscape
- Analysis and gap identification
- `web_fetch` on primary sources
- Synthesis and assumption challenging
- Repeat for depth
**Required:** Explicit analysis between every tool call showing evolution of understanding.
### Phase 4: Report
Academic narrative with:
- Executive Summary
- Knowledge Development
- Comprehensive Analysis
- Practical Implications
- APA References
## File Structure
```
deep-research/
├── SKILL.md # Full methodology (500+ lines)
├── README.md # This file
├── quickref.md # One-page cheat sheet
├── example.md # Complete workflow example
└── LICENSE # Apache 2.0
```
## When to Use This
- Literature reviews requiring academic rigor
- Competitive intelligence with source verification
- Complex topics needing multi-source synthesis
- Any research where you need to **show your work**
- When you don't trust black-box AI summaries
## License
Apache 2.0 — See [LICENSE](LICENSE)
---
**Built for researchers who care about methodology, not just outputs.**