fireworks-tech-graph

---
name: fireworks-tech-graph
description: Generate production-quality SVG+PNG technical diagrams from natural language using Claude Code. Supports 8 diagram types, 5 visual styles, and deep AI/Agent domain knowledge.
triggers:
  - generate a diagram
  - draw an architecture diagram
  - create a technical diagram
  - visualize my system
  - make a flowchart
  - draw a sequence diagram
  - generate SVG diagram
  - create architecture visualization
---

# fireworks-tech-graph

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

Turn natural language descriptions into polished SVG + PNG technical diagrams. Supports 8 diagram types, 5 visual styles, and deep knowledge of AI/Agent system patterns (RAG, Mem0, Multi-Agent, Tool Call flows).

---

## Installation

```bash
# Via Claude Code skills
claude skills install fireworks-tech-graph

# Or clone directly
git clone https://github.com/yizhiyanhua-ai/fireworks-tech-graph.git ~/.claude/skills/fireworks-tech-graph

System dependency (required for PNG export)

# macOS
brew install librsvg

# Ubuntu/Debian
sudo apt install librsvg2-bin

# Verify
rsvg-convert --version

---

How It Works

User prompt → Skill classifies diagram type + style
           → Generates SVG with semantic shapes + arrows
           → Runs: rsvg-convert -w 1920 input.svg -o output.png
           → Reports file paths of .svg and .png

Output files are written to the current directory (or --output path).

---

Basic Usage

Trigger phrases (auto-detected)

generate diagram / draw diagram / create chart / visualize
architecture diagram / flowchart / sequence diagram / data flow

Simple requests

Draw a RAG pipeline flowchart
Generate an Agentic Search architecture diagram
Create a tool call flow diagram
Visualize a microservices architecture

Specify style

Draw a microservices architecture diagram, style 2 (dark terminal)
Draw a multi-agent collaboration diagram --style glassmorphism
Generate a Mem0 architecture diagram, blueprint style

Specify output path

Generate a Mem0 architecture diagram, output to ~/Desktop/
Create a tool call flow diagram --output /tmp/diagrams/

---

Visual Styles

#	Name	Background	Font	Best For
1	Flat Icon (default)	#ffffff	Helvetica	Blogs, slides, docs
2	Dark Terminal	#0f0f1a	SF Mono / Fira Code	GitHub README, dev articles
3	Blueprint	#0a1628	Courier New	Architecture docs, engineering
4	Notion Clean	#ffffff	system-ui	Notion, Confluence, wikis
5	Glassmorphism	#0d1117 gradient	Inter	Product sites, keynotes

Reference files for each style live in references/style-N-*.md with exact color tokens and SVG patterns.

---

Diagram Types

Type	Description	Key Layout Rule
Architecture	Services, components, cloud infra	Horizontal layers top→bottom
Data Flow	What data moves where	Label every arrow with data type
Flowchart	Decisions, process steps	Diamond = decision, top→bottom
Agent Architecture	LLM + tools + memory	5-layer model: Input/Agent/Memory/Tool/Output
Memory Architecture	Mem0, MemGPT-style	Separate read/write paths, memory tiers
Sequence	API call chains, time-ordered	Vertical lifelines, horizontal messages
Comparison	Feature matrix, side-by-side	Column = system, row = attribute
Mind Map	Concept maps, radial	Central node, bezier branches

---

AI/Agent Domain Patterns (Built-in)

The skill has pre-loaded knowledge of these patterns — just name them:

RAG Pipeline         → Query → Embed → VectorSearch → Retrieve → LLM → Response
Agentic RAG          → RAG + Agent loop + Tool use
Agentic Search       → Query → Planner → [Search/Calc/Code] → Synthesizer
Mem0 Memory Layer    → Input → Memory Manager → [VectorDB + GraphDB] → Context
Agent Memory Types   → Sensory → Working → Episodic → Semantic → Procedural
Multi-Agent          → Orchestrator → [SubAgent×N] → Aggregator → Output
Tool Call Flow       → LLM → Tool Selector → Execution → Parser → LLM (loop)

Example prompts for each:

Generate a Mem0 memory architecture diagram with vector store, graph DB, KV store, and memory manager
Draw a Multi-Agent diagram: Orchestrator dispatches 3 SubAgents (search / compute / code), results aggregated
Visualize the Tool Call execution flow: LLM → Tool Selector → Execution → Parser → back to LLM
Compare Agentic RAG vs standard RAG in a feature matrix, Notion clean style
Draw the 5 agent memory types: Sensory, Working, Episodic, Semantic, Procedural

---

Shape Vocabulary

Shapes carry semantic meaning consistently across all styles:

Concept	Shape
User / Human	Circle + body
LLM / Model	Rounded rect, double border, ⚡
Agent / Orchestrator	Hexagon
Memory (short-term)	Dashed-border rounded rect
Memory (long-term)	Solid cylinder
Vector Store	Cylinder with inner rings
Graph DB	3-circle cluster
Tool / Function	Rect with ⚙
API / Gateway	Hexagon (single border)
Queue / Stream	Horizontal pipe/tube
Document / File	Folded-corner rect
Browser / UI	Rect with 3-dot titlebar
Decision	Diamond
External Service	Dashed-border rect

---

Arrow Semantics

Flow Type	Stroke	Dash	Meaning
Primary data flow	2px solid	—	Main request/response
Control / trigger	1.5px solid	—	System A triggers B
Memory read	1.5px solid	—	Retrieve from store
Memory write	1.5px	5,3	Write/store operation
Async / event	1.5px	4,2	Non-blocking
Feedback / loop	1.5px curved	—	Iterative reasoning

---

Example Prompts by Scenario

AI/Agent Systems

Draw a microservices architecture: Client → API Gateway → [User Service / Order Service / Payment Service] → PostgreSQL + Redis
Generate a data pipeline: Kafka → Spark → S3 → Athena, blueprint style
Draw a Kubernetes deployment: Ingress → Service → [Pod × 3] → ConfigMap + PersistentVolume

API & Sequence Flows

Draw an OAuth2 authorization code flow sequence diagram: User → Client → Auth Server → Resource Server
Draw the ChatGPT Plugin call sequence diagram

Decision & Process Flows

Draw a pre-launch QA flowchart: Code Review → Security Scan → Performance Test → Manual Approval → Deploy
Generate a feature comparison matrix: RAG vs Fine-tuning vs Prompt Engineering

Concept Maps

Visualize the LLM application tech stack: foundation model → SDK → app framework → deployment
Draw an AI Agent capability map: Perception / Memory / Reasoning / Action / Learning

---

SVG Generation Guidelines

When generating SVG diagrams, follow these rules:

Canvas & Layout

viewBox: "0 0 1600 900"   ← 16:9 default
preserveAspectRatio: "xMidYMid meet"
Padding: 60px all sides
Layer spacing: 140px vertical between swim lanes
Node spacing: 180px horizontal minimum

Style 1 — Flat Icon (default) key tokens

<!-- Background -->
<rect width="1600" height="900" fill="#ffffff"/>

<!-- Node: LLM (double border) -->
<rect x="200" y="200" width="160" height="60" rx="8"
      fill="#EEF2FF" stroke="#6366F1" stroke-width="2"/>
<rect x="204" y="204" width="152" height="52" rx="6"
      fill="none" stroke="#6366F1" stroke-width="1" opacity="0.5"/>

<!-- Node: Agent (hexagon) -->
<polygon points="380,200 420,180 460,200 460,240 420,260 380,240"
         fill="#F0FDF4" stroke="#22C55E" stroke-width="2"/>

<!-- Primary arrow -->
<line x1="360" y1="230" x2="380" y2="230"
      stroke="#6366F1" stroke-width="2" marker-end="url(#arrowhead)"/>

<!-- Async arrow (dashed) -->
<line x1="360" y1="230" x2="380" y2="230"
      stroke="#94A3B8" stroke-width="1.5" stroke-dasharray="4,2"
      marker-end="url(#arrowhead-gray)"/>

Style 2 — Dark Terminal key tokens

<rect width="1600" height="900" fill="#0f0f1a"/>
<!-- Nodes use: fill="#1a1a2e", stroke="#00ff88" or "#ff6b6b" or "#4ecdc4" -->
<!-- Text: fill="#e2e8f0", font-family="'SF Mono', 'Fira Code', monospace" -->
<!-- Primary arrows: stroke="#00ff88" -->

Style 3 — Blueprint key tokens

<rect width="1600" height="900" fill="#0a1628"/>
<!-- Grid lines: stroke="#1e3a5f" stroke-width="1" opacity="0.4" -->
<!-- Nodes: fill="#0d2137", stroke="#00bcd4" stroke-width="1.5" -->
<!-- Text: fill="#b0c4de", font-family="'Courier New', monospace" -->
<!-- Arrows: stroke="#00bcd4" -->

Style 5 — Glassmorphism key tokens

<defs>
  <linearGradient id="bg" x1="0%" y1="0%" x2="100%" y2="100%">
    <stop offset="0%" style="stop-color:#0d1117"/>
    <stop offset="100%" style="stop-color:#161b22"/>
  </linearGradient>
  <filter id="blur"><feGaussianBlur stdDeviation="4"/></filter>
</defs>
<rect width="1600" height="900" fill="url(#bg)"/>
<!-- Glass card: fill="rgba(255,255,255,0.05)", stroke="rgba(255,255,255,0.1)"
     backdrop blur via filter, border-radius 12px -->

Arrowhead defs (include in every SVG)

<defs>
  <marker id="arrowhead" markerWidth="10" markerHeight="7"
          refX="9" refY="3.5" orient="auto">
    <polygon points="0 0, 10 3.5, 0 7" fill="#6366F1"/>
  </marker>
  <marker id="arrowhead-gray" markerWidth="10" markerHeight="7"
          refX="9" refY="3.5" orient="auto">
    <polygon points="0 0, 10 3.5, 0 7" fill="#94A3B8"/>
  </marker>
</defs>

---

PNG Export Command

After generating the SVG file, always run this to produce the PNG:

rsvg-convert -w 1920 output.svg -o output.png

For custom dimensions:

# 2× retina width
rsvg-convert -w 3840 output.svg -o output@2x.png

# Fixed height
rsvg-convert -h 1080 output.svg -o output-1080.png

# Both dimensions (may distort — use only if aspect ratio matches)
rsvg-convert -w 1920 -h 1080 output.svg -o output.png

---

File Naming Convention

{topic}-{type}-{style}.svg
{topic}-{type}-{style}.png

Examples:
  mem0-memory-architecture-dark.svg
  mem0-memory-architecture-dark.png
  rag-pipeline-flowchart-flat.svg
  multi-agent-architecture-glass.png

---

Project File Structure

fireworks-tech-graph/
├── SKILL.md                      # Main skill definition
├── README.md                     # English docs
├── README.zh.md                  # Chinese docs
├── references/
│   ├── style-1-flat-icon.md      # Color tokens, SVG patterns
│   ├── style-2-dark-terminal.md
│   ├── style-3-blueprint.md
│   ├── style-4-notion-clean.md
│   ├── style-5-glassmorphism.md
│   └── icons.md                  # 40+ product icons + semantic shapes
└── assets/
    └── samples/                  # Sample PNG outputs per style

---

Product Icon Coverage

When a node represents a known product, use its brand color as the node accent:

Product	Brand Color
OpenAI	#00A67E
Anthropic/Claude	#D97757
Google Gemini	#4285F4
Meta LLaMA	#0668E1
Pinecone	#1A1A2E + #00FF88
Weaviate	#FA0050
LangChain	#1C3C3C
Kafka	#231F20 + #FF6B35
PostgreSQL	#336791
Redis	#DC382D
Kubernetes	#326CE5
AWS	#FF9900
GCP	#4285F4
Azure	#0078D4

Full icon reference: references/icons.md

---

Troubleshooting

rsvg-convert: command not found

# macOS — install librsvg
brew install librsvg

# Ubuntu
sudo apt-get update && sudo apt-get install -y librsvg2-bin

# Check PATH
which rsvg-convert

PNG is blank or all black

• Check SVG has explicit width/height or a valid viewBox
• Ensure fill colors are not transparent on the root <rect>
• Dark style backgrounds need explicit background rect before other elements

Text appears as boxes / missing glyphs

• The skill uses only system-safe fonts: Helvetica, Arial, Courier New, system-ui, monospace
• Avoid Google Fonts or web fonts — rsvg-convert does not fetch external resources
• For emoji icons (⚡ ⚙), test with rsvg-convert — fall back to geometric SVG shapes if unsupported

SVG renders fine but PNG is low resolution

# Increase width — 1920 is minimum, 3840 for retina
rsvg-convert -w 3840 input.svg -o output@2x.png

Diagram is too cramped

• Increase canvas: change viewBox="0 0 1600 900" to "0 0 2400 1200"
• Increase node spacing: use 220px horizontal, 160px vertical between layers
• Add swim lane padding: 40px inside each lane boundary

Style not applying

• Reference the style by number (1–5) or name: flat, dark, blueprint, notion, glass
• If ambiguous, default is Style 1 (Flat Icon)

---

Quick Reference Card

STYLES:    1=flat  2=dark  3=blueprint  4=notion  5=glass
TYPES:     architecture | dataflow | flowchart | agent
           memory | sequence | comparison | mindmap
SHAPES:    circle=user  hexagon=agent  double-rect=LLM
           cylinder=storage  diamond=decision  pipe=queue
ARROWS:    solid=primary  dashed(5,3)=write  dashed(4,2)=async
EXPORT:    rsvg-convert -w 1920 input.svg -o output.png
CANVAS:    viewBox="0 0 1600 900"  padding=60px