メニュー
Extract structured CFD information from research papers using LLM-powered template filling. Analyze combustion simulation methodologies, turbulence models, and solver settings.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Set up DeepFlame combustion cases with chemistry, boundary conditions, and solver configuration. Use for combustion simulations involving dfLowMachFoam, dfSprayFoam, dfHighSpeedFoam.
Set up complete OpenFOAM CFD cases with proper boundary conditions, solver configuration, and physics modeling. Use for non-reacting flows, incompressible flows, and general CFD simulations.
| name | Paper Analysis |
| description | Extract structured CFD information from research papers using LLM-powered template filling. Analyze combustion simulation methodologies, turbulence models, and solver settings. |
NEVER use the read tool on PDF files. Always convert PDFs to markdown first:
Convert PDF to Markdown:
python skills/paper_analysis/scripts/convert_pdf_to_md.py \
--input_pdf /absolute/path/to/paper.pdf \
--output_dir /absolute/path/to/output \
--method markitdown
Then analyze the converted .md file using paper_analysis_tool
# Extract CFD data from converted paper
result = paper_analysis_tool(
action="extract",
markdown_path="/path/to/converted_paper.md",
template_path="skills/paper_analysis/templates/default_template.yaml",
output_path="/path/to/extracted_data.json"
)
# List available templates
templates = paper_analysis_tool(action="list_templates")
# Validate extraction results
validation = paper_analysis_tool(
action="validate",
path="/path/to/extracted_data.json"
)
pip install markitdownpython skills/paper_analysis/scripts/convert_pdf_to_md.py \
--input_pdf /absolute/path/to/paper.pdf \
--output_dir /absolute/path/to/output \
--method markitdown
pip install magic-pdfpython skills/paper_analysis/scripts/convert_pdf_to_md.py \
--input_pdf /absolute/path/to/paper.pdf \
--output_dir /absolute/path/to/output \
--method mineru
Output Structure:
output_dir/
└── paper_name/
└── auto/
├── paper_name.md # Converted markdown
└── [other files...] # Images, JSON (MinerU only)
from dflagentic.agents.tools.paper_analysis_tool import paper_analysis_tool
result = paper_analysis_tool(
action="extract",
markdown_path="/path/to/paper.md",
template_path="/path/to/template.yaml",
output_path="/path/to/output.yaml",
output_format="yaml"
)
result = paper_analysis_tool(
action="validate_template",
template_path="/path/to/template.yaml"
)
result = paper_analysis_tool(action="list_templates")
The skill uses a sophisticated template system with contextual prompts:
turbulence_modeling:
rans_les_approach:
type: "enum"
required: true
options: ["RANS", "LES", "DNS", "DES", "SAS", "URANS"]
description: "Turbulence modeling approach"
prompt: |
Determine the turbulence modeling approach used.
- RANS: Reynolds-Averaged Navier-Stokes (most common for engineering)
- LES: Large Eddy Simulation (high fidelity, expensive)
- DNS: Direct Numerical Simulation (highest fidelity, very expensive)
Look for explicit mentions or infer from context like mesh resolution,
computational cost discussions, or subgrid modeling.
OPENAI_API_KEY: Standard OpenAI API keyAZURE_API_KEY: Azure OpenAI API keyAZURE_API_BASE: Azure endpoint URLAZURE_API_VERSION: Azure API versionskills/paper_analysis/templates/skills_tool load paper_analysispaper_analysis_tool in the agent tools systemdf-agent paper commandsThis skill provides powerful LLM-powered paper analysis capabilities integrated seamlessly into the DeepFlame agent system.