Skip to content
Conduit

skill

tech-diagrams

Agent Skill: Generate publication-quality technical diagrams using Gemini image generation. Works with Claude Code, Cursor, Copilot, Windsurf, Continue, Aider, and more.

smwitkowski/tech-diagramscommunitylistedunreviewed

Set up

Skills are installed by placing their folder where your agent loads skills.

The download comes from github.com/smwitkowski/tech-diagrams itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 26, 2026

tech-diagrams

Generate publication-quality technical diagrams using Gemini image generation. Works across Claude Code, Cursor, Copilot, Windsurf, Continue, Aider, Roo Code, and any tool that supports Agent Skills.

What it does

Takes a natural language description of a system and produces a polished, publication-ready PNG diagram. The skill reads your codebase and architecture docs, selects the right diagram type and visual style, composes a detailed prompt with spatial layout instructions, auto-injects matching brand icons from a 536-icon library, and generates the image via Gemini's native image generation.

Supported diagram types

  • C1 System Context — system + external actors/systems
  • C2 Container — internal deployable units grouped by concern
  • C3 Component — internals of a single container
  • Network & Security — access paths, trust boundaries, private endpoints
  • Data Flow — end-to-end request lifecycle with numbered steps
  • Sequence — temporal orchestration between components
  • Use Case Map — feature/capability matrix
  • Comparison / Split-Screen — before/after, current/future, option A vs. B
  • Layered Ecosystem Map — product positioning within a technology stack
  • Cyclical / Measurement Loop — continuous process cycles (DevOps, FinOps)
  • Decision Tree / Flowchart — question-driven selection flows
  • Deployment Topology — where software runs on infrastructure
  • Concentric Rings / Containment — layered protection or abstraction rings
  • Threat Model / Attack Surface — attack vectors and mitigations

Prerequisites

RequirementDetails
ADC / Vertex (recommended)GOOGLE_GENAI_USE_VERTEXAI=1, GOOGLE_CLOUD_PROJECT — no API key on GCP
or GEMINI_API_KEYLocal dev via AI Studio
python3scripts/build-request.py, scripts/generate.py

Installation

Claude Code

Clone into your skills directory:

git clone https://github.com/smwitkowski/tech-diagrams.git ~/.claude/skills/tech-diagrams

The skill auto-registers via SKILL.md and appears as /tech-diagrams.

Cursor

Clone the repo into your project (or as a submodule), then the .cursor/rules/tech-diagrams.mdc adapter loads automatically when relevant.

git clone https://github.com/smwitkowski/tech-diagrams.git .cursor/tech-diagrams

Or copy .cursor/rules/tech-diagrams.mdc into your project's .cursor/rules/ directory and adjust paths.

GitHub Copilot

The .github/instructions/tech-diagrams.instructions.md file activates when working with diagram-related files. Clone the repo and copy the instructions file:

cp tech-diagrams/.github/instructions/tech-diagrams.instructions.md your-project/.github/instructions/

Windsurf

The .windsurf/rules/tech-diagrams.md adapter activates on model decision. Clone and copy:

cp tech-diagrams/.windsurf/rules/tech-diagrams.md your-project/.windsurf/rules/

Continue.dev

Copy the prompt file into your Continue configuration:

cp tech-diagrams/.continue/prompts/tech-diagrams.md ~/.continue/prompts/

Then invoke with @tech-diagrams in Continue.

Aider

Add to your .aider.conf.yml:

read: path/to/tech-diagrams/SKILL.md

Roo Code / Generic Agent Skills

Any tool that reads SKILL.md or AGENTS.md at the repo root will pick up the skill automatically. Clone the repo and point your tool at it.

Examples

All examples generated with gemini-3.1-flash-image-preview. Each prompt is in the corresponding .prompt.txt file in examples/.

Network & Security — GCP Enterprise Landing Zone

Hub-and-spoke VPC architecture with Cloud Armor, IAP, VPN, and workload VPCs across three trust zones.

Comparison — Monolith to Microservices Migration

Split-screen showing on-premises Java/Oracle monolith vs. GCP Cloud Run microservices with Cloud SQL and Pub/Sub.

Layered Ecosystem — AI/ML Platform Stack

Five-layer technology stack from infrastructure to applications, with Vertex AI, BigQuery, GKE, and cross-cutting security/observability sidebars.

Usage

Ask your AI coding assistant to create a diagram:

"Create a C2 container diagram of our payment processing system"

The skill will:

  1. Read your codebase to extract component names and relationships
  2. Select the appropriate diagram type and visual style
  3. Compose a detailed prompt with ASCII layout and spatial instructions
  4. Auto-inject matching brand icons from the 536-icon library
  5. Generate a publication-quality PNG via Gemini

Output is saved adjacent to the source document with a descriptive filename like payment-c2-container.png.

Icon library

The icons/ directory contains 536 verified brand icons across four providers:

ProviderDirectoryExamples
Google Cloudicons/gcp/Cloud Run, BigQuery, Pub/Sub
Azureicons/azure/App Service, Cosmos DB, Event Hub
AWSicons/aws/Lambda, DynamoDB, SQS
Material Designicons/google/material/Universal fallbacks (database, API, user)

Icons are automatically matched and injected into Gemini requests by scripts/build-request.py, ensuring consistent brand logos across all generated diagrams.

Gemini Enterprise remote agent (internal)

Hosting for the Gemini Enterprise A2A remote agent (Cloud Run, Terraform, CI) lives in the private 66degrees repo — not in this public skill tree:

github.com/66degrees/tech-diagrams-agent (66degrees engineers only).

This public repository remains the skill + icon library consumed by that agent at a pinned git ref.

License

MIT

↑↓ navigate openesc close