Fixture-Graph Skill

Fixture-Graph Skill

⚡ Quick Start for Agents

Don't get overwhelmed by 50+ commands! Use domain navigation:

# Step 1: Find your domain
fixture-graph domains

# Step 2: List commands for your domain
fixture-graph list --domain ml        # ML/​LLM projects
fixture-graph list --domain control   # Aerospace/​control systems
fixture-graph list --domain bio       # Bioinformatics

# Step 3: Or get recommendations by data type
fixture-graph recommend --data-type classification
fixture-graph recommend --data-type time_series
fixture-graph recommend --show-types  # See all data types

Domain Quick Reference

---

Generate publication-quality figures from code analysis data for academic papers.

Architecture

Multi-backend design for maximum compatibility:

D3-Style Scientific Visualizations

Advanced visualization types for scientific and engineering papers:

Control Systems & Engineering Plots

Specialized visualizations for aerospace, nuclear, and control systems:

GPU/​Hardware Performance Visualizations

Specialized visualizations for GPU, CUDA, and hardware performance papers:

LLM/​ML Visualization Types

Specialized visualizations for machine learning and LLM papers:

Biology/​Bioinformatics Visualizations

Specialized visualizations for biological and medical research:

Commands

deps - Dependency Graph

Generate dependency graph from Python project.

./​run.sh deps --project /​path/​to/​package --output deps.pdf
./​run.sh deps -p ./​src -o deps.svg --backend mermaid --depth 3
./​run.sh deps -p ./​src -o deps.json --backend networkx --format json

Options:

• --project, -p: Path to Python package/​module (required)
• --output, -o: Output file (default: dependencies.pdf)
• --format, -f: Output format (pdf, png, svg, dot, json)
• --depth, -d: Maximum dependency depth (default: 2)
• --backend, -b: graphviz, mermaid, networkx

uml - UML Class Diagram

Generate UML class diagram using pyreverse.

./​run.sh uml --project ./​src --output classes.pdf

Requires: pip install pylint

architecture - Architecture Diagram

Generate architecture diagram from project or /​assess JSON.

./​run.sh architecture --project ./​assess_output.json --output arch.pdf
./​run.sh architecture -p /​path/​to/​project -o arch.svg --backend mermaid

metrics - Metrics Chart

Generate publication-quality metrics chart with IEEE styling.

./​run.sh metrics --input data.json --output metrics.pdf --type bar
./​run.sh metrics -i data.json -o chart.pdf --type pie --title "Issue Distribution"

Chart types: bar, hbar, pie, line

Input formats:

// Simple dict
{"Feature A": 42, "Feature B": 28}

// With "metrics" key
{"metrics": {"LOC": 1500, "Functions": 45}}

// List format
[{"name": "A", "value": 10}, {"name": "B", "value": 20}]

table - LaTeX Table

Generate LaTeX table with proper escaping.

./​run.sh table --input features.json --output table.tex --caption "Feature Comparison"

workflow - Workflow Diagram

Generate workflow/​pipeline diagram with quality gates.

./​run.sh workflow --stages "Scope,Analysis,Search,Learn,Draft" --output workflow.pdf
./​run.sh workflow -s "A,B,C,D" -o flow.svg --no-gates --backend graphviz

theorem - Formal Verification

Generate formally verified theorem from requirement (uses lean4-prove).

./​run.sh theorem --requirement "All inputs must be validated" --name input_validation --output theorem.lean

sankey - Sankey Diagram

Generate Sankey diagram for energy/​mass flow balances.

./​run.sh sankey --input flows.json --output sankey.pdf --title "Reactor Coolant Flow"

Input format:

[{"source": "Primary", "target": "Heat Exchanger", "value": 1000}]

heatmap - Heatmap

Generate heatmap for field distributions or correlation matrices.

./​run.sh heatmap --input matrix.json --output flux.pdf --cmap plasma --title "Neutron Flux"

treemap - Treemap

Generate treemap for hierarchical size data.

./​run.sh treemap --input sizes.json --output breakdown.pdf

sunburst - Sunburst Chart

Generate sunburst chart for hierarchical fault trees.

./​run.sh sunburst --input hierarchy.json --output faults.pdf

force-graph - Force-Directed Graph

Generate force-directed graph for system topology.

./​run.sh force-graph --input network.json --output topology.pdf

parallel-coords - Parallel Coordinates

Generate parallel coordinates for multi-dimensional DOE analysis.

./​run.sh parallel-coords --input experiments.json --output doe.pdf --color-by efficiency

radar - Radar Chart

Generate radar/​spider chart for multi-attribute comparison.

./​run.sh radar --input attributes.json --output comparison.pdf

Input format:

{"Design A": {"Safety": 8, "Cost": 6, "Efficiency": 9}, "Design B": {"Safety": 9, "Cost": 4, "Efficiency": 7}}

bode - Bode Plot

Generate Bode plot for control systems frequency response.

./​run.sh bode --num 1,2 --den 1,3,2 --output bode.pdf --freq-min 0.01 --freq-max 100

nyquist - Nyquist Plot

Generate Nyquist plot for stability analysis.

./​run.sh nyquist --num 1,2 --den 1,3,2,0 --output nyquist.pdf

rootlocus - Root Locus

Generate root locus for control system gain analysis.

./​run.sh rootlocus --num 1 --den 1,5,6 --output rootlocus.pdf

polar - Polar Plot

Generate polar plot for directional data (antenna patterns, wind roses).

./​run.sh polar --input pattern.json --output antenna.pdf

contour - Contour Plot

Generate contour plot for field distributions (flux, temperature, stress).

./​run.sh contour --input field.json --output flux.pdf --cmap plasma --levels 30

Input format:

{"x": [0, 1, 2], "y": [0, 1, 2], "z": [[0,1,2],[1,2,3],[2,3,4]]}

gantt - Gantt Chart

Generate Gantt chart for project scheduling.

./​run.sh gantt --input schedule.json --output timeline.pdf

Input format:

[{"task": "Design", "start": 0, "end": 5, "progress": 100}]

pert - PERT Network

Generate PERT network diagram for critical path analysis.

./​run.sh pert --input network.json --output cpm.pdf

vector-field - Vector Field

Generate vector field for flow visualization.

./​run.sh vector-field --input flow.json --output velocity.pdf --streamlines

phase-portrait - Phase Portrait

Generate phase portrait for dynamical systems.

./​run.sh phase-portrait --equations "dx = y; dy = -x - 0.5*y" --output damped.pdf

roofline - Roofline Plot

Generate roofline plot for GPU/​hardware performance analysis.

./​run.sh roofline --input kernels.json --output roofline.pdf

Input format:

{"peak_flops": 19.5e12, "peak_bandwidth": 900e9, "kernels": [{"name": "GEMM", "flops": 1e12, "bytes": 1e9}]}

scaling-law - Scaling Law Plot

Generate log-log scaling law plot (common in LLM papers).

./​run.sh scaling-law --input params_vs_loss.json --output scaling.pdf --fit

confusion-matrix - Confusion Matrix

Generate confusion matrix for classification results.

./​run.sh confusion-matrix --input results.json --output confusion.pdf --normalize

roc-curve - ROC Curve

Generate ROC curve with AUC for binary classification.

./​run.sh roc-curve --input roc_data.json --output roc.pdf

training-curves - Training Curves

Generate multi-run training curves with std shading.

./​run.sh training-curves --input runs.json --output loss.pdf --log-y

attention-heatmap - Attention Heatmap

Generate transformer attention visualization.

./​run.sh attention-heatmap --input attention.json --output attn.pdf

embedding-scatter - Embedding Scatter

Generate t-SNE or UMAP visualization of embeddings.

./​run.sh embedding-scatter --input embeddings.json --output tsne.pdf --method tsne

throughput-latency - Throughput vs Latency

Generate throughput vs latency plot for inference benchmarks.

./​run.sh throughput-latency --input benchmarks.json --output perf.pdf

pr-curve - Precision-Recall Curve

Generate Precision-Recall curve for classification.

./​run.sh pr-curve --input pr_data.json --output pr.pdf

violin - Violin Plot

Generate violin plot for distribution comparison.

./​run.sh violin --input expression.json --output violin.pdf --y-label "Expression"

volcano - Volcano Plot

Generate volcano plot for differential expression analysis.

./​run.sh volcano --input deseq.json --output volcano.pdf --fc 1.5 --pval 0.01

survival-curve - Kaplan-Meier Survival Curve

Generate Kaplan-Meier survival curve.

./​run.sh survival-curve --input survival.json --output km.pdf

manhattan - Manhattan Plot

Generate Manhattan plot for GWAS results.

./​run.sh manhattan --input gwas.json --output manhattan.pdf

feature-importance - Feature Importance

Generate feature importance bar chart.

./​run.sh feature-importance --input importance.json --output features.pdf --top-n 15

calibration - Calibration Plot

Generate calibration plot (reliability diagram).

./​run.sh calibration --input calibration.json --output calib.pdf

from-assess - Generate All Figures

Generate all figures from /​assess output in one command.

./​run.sh from-assess --input assess_output.json --output-dir ./​figures/

Generates:

• architecture.pdf - System architecture diagram
• dependencies.pdf - Module dependency graph
• features.pdf - Feature distribution chart
• issues.pdf - Issue severity pie chart
• comparison.tex - Feature comparison table
• test_coverage.tex - Test coverage table

check - Backend Status

Check which backends are available.

./​run.sh check

Publication Quality Settings

matplotlib figures use IEEE publication settings:

• Font: 8pt Times New Roman (serif)
• DPI: 600 for saving, 300 for display
• Column widths: Single (3.5"), Double (7.16")
• TrueType fonts for Illustrator compatibility

Integration with paper-writer

from pathlib import Path
import subprocess

def generate_figures(assess_json: Path, output_dir: Path):
    subprocess.run([
        str(FIXTURE_GRAPH_SCRIPT), "from-assess",
        "--input", str(assess_json),
        "--output-dir", str(output_dir),
    ])

Dependencies

Required:

• Python 3.10+
• typer
• numpy

Optional (enables features):

System dependencies:

• graphviz (apt install graphviz) - Graphviz rendering
• mermaid-cli (npm install -g @mermaid-js/​mermaid-cli) - Mermaid backend
• lean4-prove skill - Formal theorem verification

Installation

# Core
pip install typer numpy matplotlib

# Full installation (all features)
pip install typer numpy matplotlib seaborn plotly networkx pandas squarify scipy control pydeps pylint

# Control systems only
pip install typer numpy matplotlib scipy control

# System dependencies
apt install graphviz  # Debian/​Ubuntu
npm install -g @mermaid-js/​mermaid-cli

Sanity Check

./​sanity.sh

Verifies:

• Python dependencies available
• CLI loads correctly
• Basic diagram generation works
• Tests pass

Research Sources

This skill's design is informed by:

• pydeps - Python module dependency visualization
• matplotlib for papers - Publication-quality figures
• Mermaid vs Graphviz - Diagram-as-code comparison
• NetworkX + D3 - Interactive graph visualization
• OverViz - AI architecture diagram generation