review-analysis

name: review_analysis description: Research assistant for extracting detailed information from review articles (002). Produces citation-ready content with bib-key lookups for thesis incorporation.

Thesis Review Article Analysis

Extracts detailed information from review articles (002) sources) for incorporation into the thesis text. Unlike literature_research (which identifies relevant reviews), this skill extracts and structures content from specific review articles for direct use in writing.

Target Notebook

• Name: thesis references
• ID: 1b7df790-7858-4fc8-879c-39f41238c4ae
• Strict Rule: Exclusively use this notebook.

When to Use This Skill

Use this skill when:

• Incorporating review content into thesis text — the review is NOT included directly in the thesis, so its information must be paraphrased and cited
• Extracting the narrative structure of a review (e.g., "How does Cirelli (2024) structure the argument for indirect detection?")
• Getting precise definitions or derivations from review articles
• Understanding how a topic transitions from one concept to another in expert exposition

Do NOT use this skill for general reference gathering — use literature_research instead. Do NOT use this skill for the author's own papers — use paper_analysis instead.

Key Principle

Review articles are external to the thesis. Their content must be:

1. Paraphrased — never copy text directly
2. Cited — every claim needs a \cite{bib_key} reference
3. Integrated — woven into the thesis narrative, not just summarized

Prerequisites

1. Run source_registry first to obtain 002) source IDs.
2. Typically called after literature_research has identified which reviews are most relevant.

MCP Tool Usage

Querying Specific Reviews

mcp_notebooklm_notebook_query(
    notebook_id="1b7df790-7858-4fc8-879c-39f41238c4ae",
    query="<your question>",
    source_ids=<review_002_ids>,  # from source_registry, only 002) sources
    conversation_id=<previous_id>  # for follow-up drilling
)

Filtering strategy: Pass only the 002) review source IDs to focus the AI on review literature. If you need a specific review, pass only that single source ID.

Targeted Single-Source Queries

When you need to extract from a specific review article:

mcp_notebooklm_notebook_query(
    notebook_id="1b7df790-7858-4fc8-879c-39f41238c4ae",
    query="Based on this review, detail the structure and key arguments about [topic]",
    source_ids=["<single_review_source_id>"]
)

Query Strategy

1. Structure Extraction: Ask for the table of contents or argument flow of a specific review.

   • Prompt: "What is the detailed structure of [Review X] regarding [Topic]? List section headers and summarize the argumentation flow."

2. Concept Definitions: Extract how the review defines standard concepts.

   • Prompt: "How does [Review X] define [Concept]? Include any key equations or parameters."

3. Narrative Flow: Understand how experts transition between ideas.

   • Prompt: "How does [Review X] transition from [Concept A] to [Concept B]? What logical steps connect them?"

4. Specific Claims with Citations: Get precise statements that can be paraphrased.

   • Prompt: "What specific claims does [Review X] make about [Topic]? For each, note the section number and any papers they cite."

5. Comparative Views: Compare how different reviews cover the same topic.

   • Prompt: "Compare how [Review A] and [Review B] present [Topic]. What differences in emphasis or interpretation exist?"

6. Figure Identification: Identify key figures from papers cited in the review that illustrate core concepts.

   • Prompt: "Which figures from the papers discussed in [Review X] are considered canonical illustrations of [Topic]? For each, provide the paper's arXiv ID, figure number, and a brief description of what it shows."
   • This feeds into section_drafting Step 4b and paper_lookup Recipe 3 for actual downloads.

Output

Citation-Ready Notes

Structure extracted content as:

## [Topic] — from [Review Author (Year)]

### Key Points
- [Paraphrased claim 1] → `\cite{bib_key}`, Sec X.X
- [Paraphrased claim 2] → `\cite{bib_key}`, Sec Y.Y

### Definitions
- **[Term]**: [definition as given in review] → `\cite{bib_key}`

### Narrative Structure
1. The review begins by establishing...
2. It then transitions to...
3. The argument concludes with...

### Key Figures
- **Fig. N** from [arXiv:XXXX.XXXXX]: [description of what it shows] → illustrates [concept]
- **Fig. M** from [arXiv:YYYY.YYYYY]: [description] → illustrates [concept]

Bib Key Lookup

For every reference extracted:

1. Get the arXiv number from the review's citations
2. Search bibliography.bib: grep_search(query="<arxiv_number>", SearchPath="bibliography.bib")
3. If found → use the bib key for \cite{}
4. If NOT found → flag as N/A and note the full reference so the user can add it

Knowledge Saving

REQUIRED: Use the knowledge skill (save mode) to persist extracted insights to .agent/knowledge/. The knowledge skill defines the standard file format and handles deduplication.

Usage Examples

Scenario: Setting the Stage

• User: "I need to write the section introducing the Galactic Center Excess."
• Action: Query 002) reviews to get the history, main interpretations (DM vs. Pulsars), and current controversy.
• Query: "Based on the review articles, outline the history and leading interpretations of the Galactic Center Excess. Include specific section references."

Scenario: Defining Standard Physics

• User: "Write the equations for the NFW profile."
• Action: Query reviews for the standard formula and parameters.
• Query: "Provide the mathematical definition of the NFW density profile and explain its parameters, citing the specific sections from the reviews."

Scenario: Broad Comparison

• User: "What are the typical uncertainties in this field?"
• Action: Query 002) reviews for a high-level summary.
• Query: "According to the reviews, what are the dominant systematic uncertainties in gamma-ray dark matter searches? List them with section references."