systematic-review - SKILL.md Agent Skill

name: systematic-review description: Conduct a PRISMA-compliant systematic literature review for urban BEV fast-charging rollout planning research.

Systematic Review Skill

You are conducting a systematic literature review for a PhD dissertation on urban BEV fast-charging rollout planning. Your work follows PRISMA 2020 guidelines and maintains structured, reproducible methodology.

Core Workflow

Every paper goes through a 5-stage pipeline:

1. INGEST   → Acquire paper (arXiv ID, PDF, URL)
2. CLASSIFY → Apply inclusion/exclusion criteria
3. EXTRACT  → Populate extraction schema fields
4. MAP      → Assign to dissertation gap categories
5. SYNTHESIZE → Update literature matrix and gap analysis

Stage Descriptions

1. Ingest

Receive a paper identifier (arXiv ID, DOI, PDF path, or URL). Use workflows/ingest_paper.py:

.venv\Scripts\python.exe workflows\ingest_paper.py --source <identifier>

The script downloads the paper, extracts metadata (title, authors, year, abstract), and creates:

memory/papers/<key>.pdf (if PDF downloaded)
memory/papers/<key>_metadata.json (metadata)

2. Classify

Apply inclusion/exclusion criteria from agents/paper_classifier.md. Output a classification file:

memory/papers/<key>_classification.md

If excluded, write reason and stop. If included, proceed to extraction.

3. Extract

Read the paper content (PDF or abstract) and populate the extraction schema (see skills/systematic_review/extraction_schema.yaml). Write to:

memory/papers/<key>_extraction.md

4. Map

Assign the paper to one or more dissertation gaps from memory/research_gaps.md:

Gap 1: Misaligned spatial units
Gap 2: Lack of zoning impact analysis
Gap 3: Equity and utilization separation
Gap 4: Static optimization dominance
Gap 5: Missing meso-micro integration

5. Synthesize

Run workflows/update_matrix.py to update the literature matrix:

.venv\Scripts\python.exe workflows\update_matrix.py

Then run workflows/generate_summary.py to produce the markdown summary.

PRISMA Compliance

Follow the PRISMA 2020 protocol defined in skills/systematic_review/prisma_template.md. Key requirements:

Document every decision at each pipeline stage
Record exclusion reasons precisely
Maintain flow diagram data (counts at each stage)
Do not skip stages — every included paper must complete all 5 stages

Quality Standards

Extraction must be based on direct paper evidence, not inference
Every mapped gap must cite a specific finding from the paper
Classification decisions must reference explicit inclusion/exclusion criteria
All dates, counts, and decisions logged with timestamps

Output Files

All outputs written to memory/papers/:

<key>_metadata.json — raw metadata
<key>_classification.md — inclusion decision
<key>_extraction.md — structured data extraction
<key>_summary.md — narrative summary

Where <key> = firstauthor_year (e.g., chen_2023).