Literature Review
Processes user-provided PDFs one at a time using isolated subagents to prevent context overflow, then synthesizes findings and drafts the Introduction section.
CRITICAL CONSTRAINTS
┌────────────────────────────────────────────────────────────────────┐
│ ⛔ ORCHESTRATOR MUST NEVER READ PDF FILES DIRECTLY ⛔ │
│ │
│ The orchestrator (you, running this skill) may ONLY read: │
│ - scope.md │
│ - inventory.md │
│ - notes/ethics-summary.md │
│ - notes/papers/*.md (the condensed notes, NOT PDFs) │
│ - notes/bibliography.md │
│ - notes/literature-synthesis.md │
│ │
│ If you find yourself about to use the Read tool on a .pdf file, │
│ STOP. Spawn a subagent instead. │
├────────────────────────────────────────────────────────────────────┤
│ ⛔ EACH SUBAGENT PROCESSES EXACTLY ONE PAPER ⛔ │
│ │
│ - One Task tool call = One PDF │
│ - Never pass multiple PDFs to a single subagent │
│ - Never ask a subagent to "process the remaining papers" │
│ - Wait for each subagent to complete before spawning the next │
│ │
│ WRONG: Task("Process papers/a.pdf, papers/b.pdf, papers/c.pdf") │
│ RIGHT: Task("Process papers/a.pdf") → wait → Task("Process b.pdf")│
└────────────────────────────────────────────────────────────────────┘
Architecture: Subagent Pattern
Each paper is processed by an isolated subagent to prevent context overflow:
┌─────────────────────────────────────────────────────────────────┐
│ LITERATURE REVIEW WORKFLOW │
├─────────────────────────────────────────────────────────────────┤
│ │
│ [ORCHESTRATOR] │
│ │ │
│ ├──► Read scope.md (extract research context) │
│ │ │
│ ├──► List papers/*.pdf │
│ │ │
│ ├──► For each PDF, spawn isolated subagent: │
│ │ │ │
│ │ ├──► [Subagent] Read ONE PDF │
│ │ ├──► [Subagent] Generate notes/papers/*.md │
│ │ └──► [Subagent] Exit (context cleared) │
│ │ │
│ ├──► Read all notes/papers/*.md (small files) │
│ │ │
│ ├──► Generate bibliography + synthesis │
│ │ │
│ └──► Draft Introduction │
│ │
└─────────────────────────────────────────────────────────────────┘
Why subagents?
- Each PDF can be 10-50 pages of dense content
- Processing 30 papers in one context overflows limits
- Subagents process one paper, write condensed notes, then exit
- Orchestrator only reads small markdown note files
Prerequisites
scope.mdmust exist (from scoping step)notes/ethics-summary.mdmay exist (provides study objectives and population context)papers/folder with PDF files to process- Create directories:
notes/papers/,notes/papers-library/,drafts/
Workflow Overview
[Step 1: Read scope.md for context]
│
▼
[Step 2: Inventory papers/ folder]
│
▼
[Step 3: Process each PDF via subagent] ─── One at a time
│ └── notes/papers/*.md
▼
[Step 4: Generate bibliography] ─── From all notes
│ └── notes/bibliography.md
▼
[Step 5: Synthesize literature] ─── Aggregate findings
│ └── notes/literature-synthesis.md
▼
[Step 6: Draft Introduction]
│
▼
[Output] ─── drafts/introduction.md
Step 1: Read Scope Context
From scope.md, extract comprehensive context to pass to each subagent:
- Research question (frames how to evaluate relevance)
- Key findings (what results need literature context)
- Target journal (determines citation depth)
- Hypothesis (what the manuscript argues)
- Population/methods (for identifying comparable studies)
Create a detailed scope summary to pass to each subagent:
## Manuscript Context
You are helping write a scientific research manuscript. Your job is to read ONE paper and extract information most relevant to this manuscript.
### Our Research Question
[Full research question from scope.md]
### Our Hypothesis
[What we expect to find / are arguing]
### Our Key Findings
[Summary of main results - what we found]
### Our Methods
[Brief description of study design, population, techniques used]
### Target Journal
[Journal name] - [word limit] words - [citation style]
### What to Look For in This Paper
As you read, focus on extracting:
1. **Background claims** we can cite to establish the problem's importance
2. **Methodological precedents** that justify our approach
3. **Comparable results** we can compare our findings against
4. **Contradictory findings** we need to address in Discussion
5. **Gaps identified** that our study fills
6. **Quotable statements** that frame the field well
If notes/ethics-summary.md exists, also include:
- Study objectives from ethics document
- Target population details
- Approved procedures
Step 2: Inventory Papers
List all PDFs in the papers/ folder:
ls papers/*.pdf
Create processing queue:
## Papers to Process
| # | Filename | Status |
|---|----------|--------|
| 1 | smith-2023.pdf | Pending |
| 2 | jones-2022.pdf | Pending |
| 3 | wilson-2021.pdf | Pending |
...
Copy PDFs to central library:
mkdir -p notes/papers-library
cp papers/*.pdf notes/papers-library/
Step 3: Process Each PDF via Subagent
CRITICAL: Process ONE paper at a time using the Task tool.
For each PDF in the queue:
3a. Spawn Subagent
Use the Task tool to spawn an isolated subagent.
IMPORTANT: Each Task call must specify exactly ONE PDF file. Do not batch.
Task(
subagent_type: "general-purpose",
prompt: """
⛔ YOU ARE PROCESSING EXACTLY ONE PAPER. DO NOT READ ANY OTHER PDFs. ⛔
## Your Role
You are a research assistant helping write a scientific manuscript. Your job is to:
1. Read ONE paper
2. Extract information most relevant to our manuscript
3. Write condensed notes highlighting what's useful for our paper
4. Exit immediately after
## Your ONE Paper
**PDF to read**: papers/{filename}.pdf
**Output file**: notes/papers/{filename}.md
## Manuscript Context (What We're Writing)
{scope_summary}
## Instructions
1. Read ONLY the PDF file specified above
2. Extract citation metadata (authors, title, journal, DOI, PMID)
3. Summarize the paper's main contribution
4. **Focus on relevance**: What from this paper helps our manuscript?
- Background claims we can cite?
- Methods similar to ours?
- Results we can compare against?
- Contradictions we need to address?
- Quotes that frame the problem well?
5. **Recommend inclusion**: Should this paper be cited in our manuscript?
6. Write condensed notes to the output file
7. Return a summary with your recommendation
8. EXIT - Do not process any other papers
## Note Template
# [First Author] et al., [Year]
**Processed**: [timestamp]
## Citation
- **Authors**: [full author list]
- **Title**: [title]
- **Journal**: [journal]
- **Year**: [year]
- **DOI**: [doi or "not found"]
- **PMID**: [pmid or "not found"]
## Paper Summary
[2-3 sentences on what this paper is about]
## Study Design
- **Type**: [RCT, cohort, retrospective, etc.]
- **Population**: [N, key characteristics]
- **Methods**: [key techniques used]
## Key Findings
| Finding | Statistic | Interpretation |
|---------|-----------|----------------|
| [finding] | [p-value, CI, effect size] | [what it means] |
## Relevance to Our Manuscript
**Relationship**: [Supports / Contradicts / Provides Context / Methodological Reference]
### For Introduction
- [Claims or context we can cite to frame our problem]
- [Gaps this paper identifies that we address]
### For Methods
- [Methodological precedents that justify our approach]
### For Discussion
- [Results to compare against ours]
- [Contradictions we need to explain]
## Key Quotes
> "[Quotable statement that frames the field or problem well]" (p. X)
> "[Another useful quote if available]" (p. X)
## Recommendation
**Include in draft?**: [Yes / No]
**Rationale**: [1-2 sentences - Is this paper directly relevant? Would citing it feel natural or forced?]
## Tags
#[topic1] #[topic2] #[methodology]
"""
)
3b. Wait for Completion
Wait for the subagent to complete before processing the next paper.
3c. Update Status
Track the subagent's recommendation for each paper:
| # | Filename | Status | Recommendation |
|---|----------|--------|----------------|
| 1 | smith-2023.pdf | ✓ Complete | Include |
| 2 | jones-2022.pdf | ✓ Complete | Include |
| 3 | wilson-2021.pdf | ✓ Complete | Exclude - different population |
| 4 | brown-2020.pdf | Processing... | - |
3d. Handle Errors
If a subagent fails:
- Log the error
- Mark as "Needs manual review"
- Continue with next paper
Step 4: Generate Bibliography
After ALL papers are processed, read the notes and compile bibliography.
Read each notes/papers/*.md and extract citations:
Create notes/bibliography.md:
# Master Bibliography
**Generated**: [timestamp]
**Total References**: [n]
## References
| # | Citation | DOI | PMID |
|---|----------|-----|------|
| 1 | Smith JA, et al. Title. Journal. 2023;1:1-10. | 10.xxx | 123456 |
| 2 | Jones BB, et al. Title. Journal. 2022;2:20-30. | 10.xxx | 234567 |
## Papers Needing Attention
| Paper | Issue |
|-------|-------|
| [filename] | [missing DOI / couldn't process / etc.] |
Step 5: Synthesize Literature
Read all notes/papers/*.md files and create synthesis.
Note: At this point you're only reading small markdown files, not full PDFs.
Create notes/literature-synthesis.md:
# Literature Synthesis
**Generated**: [timestamp]
**Research Question**: [from scope.md]
**Papers Reviewed**: [n]
---
## Paper Citation Tracker
Use the subagent recommendations from Step 3 to populate this table. Papers recommended for exclusion by the subagent should be marked "Not Applicable."
| # | Citation | Note File | Subagent Rec | Natural Fit | Status |
|---|----------|-----------|--------------|-------------|--------|
| 1 | Smith et al., 2023 | notes/papers/smith-2023.md | Include | Intro: establishes prevalence | Planned |
| 2 | Jones et al., 2022 | notes/papers/jones-2022.md | Include | Discussion: comparable methods | Planned |
| 3 | Wilson et al., 2021 | notes/papers/wilson-2021.md | Exclude | Different population | Not Applicable |
**Status Legend**:
- **Planned**: Subagent recommended inclusion and paper has a natural place
- **Not Applicable**: Subagent recommended exclusion or paper doesn't fit naturally
**Guidance**: Trust the subagent's assessment of relevance. The goal is a coherent manuscript, not maximum citations.
---
## Source Inventory
| # | Citation | Type | Relationship |
|---|----------|------|--------------|
| 1 | Smith et al., 2023 | Original Research | Supports |
| 2 | Jones et al., 2022 | Review | Context |
---
## Key Themes
### Theme 1: [Theme Name]
**Summary**: [2-3 sentences]
**Sources**: Smith 2023, Jones 2022
**Relevance**: [connection to our study]
### Theme 2: [Theme Name]
[repeat pattern]
---
## Supporting Findings
| Finding | Source | How It Supports Our Work |
|---------|--------|--------------------------|
| [finding] | Smith 2023 | [explanation] |
---
## Contradictory Findings
| Finding | Source | Possible Explanation |
|---------|--------|---------------------|
| [finding] | Jones 2022 | [why it differs] |
---
## Gaps in Literature
| Gap | Evidence | How We Address It |
|-----|----------|-------------------|
| [gap] | No studies on X | We examine X |
---
## Implications for Manuscript
### Introduction
- **Paragraph 1 (Problem)**: Cite [sources]
- **Paragraph 2 (Knowledge)**: Cite [sources]
- **Paragraph 3 (Gap)**: Cite [sources]
- **Paragraph 4 (This Study)**: Reference scope.md
### Discussion
- **Comparisons**: [sources for comparing results]
- **Interpretation**: [sources for mechanisms]
Step 6: Draft Introduction
Using notes/literature-synthesis.md, draft the Introduction.
Structure (4 paragraphs)
Paragraph 1: The Problem
- Significance of the clinical/scientific problem
- Epidemiology, impact, importance
Paragraph 2: Current Knowledge
- What is currently known
- Existing approaches and successes
Paragraph 3: The Gap
- What remains unknown
- Limitations of current approaches
Paragraph 4: This Study
- Purpose and approach
- What this paper contributes
Writing Guidelines
- Present tense for established facts ("MRI provides...")
- Past tense for specific findings ("Smith et al. found...")
- Every claim needs a citation [1]
- Numbered citations in order of appearance
- Aim for ~500-800 words
Draft Format
Create drafts/introduction.md:
# Introduction
[Paragraph 1 - Problem and significance]
[Paragraph 2 - Current knowledge]
[Paragraph 3 - Gap and limitations]
[Paragraph 4 - This study's purpose]
---
## References Used
1. [Citation] - notes/papers/smith-2023.md
2. [Citation] - notes/papers/jones-2022.md
...
Output
Save to:
notes/papers-library/*.pdf- Central PDF storagenotes/papers/*.md- Condensed notes per papernotes/bibliography.md- Master reference listnotes/literature-synthesis.md- Aggregated synthesisdrafts/introduction.md- Introduction draft
Return summary:
- Papers processed: [n]
- Papers recommended for inclusion: [n]
- Papers recommended for exclusion: [n]
- Papers with issues: [n]
- Key themes identified: [n]
- Introduction word count: [n]