More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Architecture Diagrams

    1

    Ask your coding agent for an architecture diagram and get one you'd actually put in a README.

    Free

    Security scanned
    Architecture Diagrams

    Architecture Diagrams

    Example session with this skill installed

    Draw my app's architecture for the README: Next.js on Vercel, Supabase for auth and the database, Stripe for payments, Resend for emails.

    • Read your context and instructions
    • Compiled the architecture diagrams

    Here's your architecture diagram. Users reach the Next.js app on Vercel; the app signs people in with Supabase Auth, stores data in Supabase Postgres, takes payments through Stripe (API calls one way, webhooks back) and sends email with Resend.

    demo-app-architecture

    What I did

    Real logos. Vercel, Next.js, Supabase, Postgres, Stripe and Resend are the official brand marks, looked up from the Simple Icons catalog. Nothing is guessed or redrawn.

    Clear layout. The request path runs left to right; Supabase and Vercel are drawn as boundaries; third-party services sit below the app.

    Checked. The layout lint found no overlaps, crossing lines or labels on borders, and I reviewed the rendered image before handing it over.

    Files

    FileWhat it is
    docs/architecture/app-architecture.svgSharp image for your README (all logos embedded)
    docs/architecture/app-architecture.drawioEditable diagram; open in draw.io or diagrams.net
    docs/architecture/app-architecture.yamlLayout spec, so later changes start here

    Add it to your README

    
    ## Architecture
    
    ![Architecture](docs/architecture/app-architecture.svg)
    
    Editable source: [app-architecture.drawio](docs/architecture/app-architecture.drawio)
    

    Next steps

    • Using Supabase Storage, Edge Functions or a cron job? Tell me and I'll add them.
    • Want it to match your repo exactly? Ask me to read the code and I'll pick up the real service names.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Ask your coding agent for an architecture diagram and get one you'd actually put in a README.

    Without this skill, agents guess icon names, stack labels on top of each other and draw arrows everywhere. This skill gives them a method:

    • Real icons, never guessed. Every icon is looked up in a catalog: 952 AWS, 363 GCP and 78 Kubernetes icons, draw.io's Azure, IBM and SAP sets, the Microsoft Fabric pack and 3,000+ brand logos. A missing icon fails loudly instead of becoming a blank box.
    • Asks about missing logos. If a product or internal app has no icon, it asks once: look up the official logo, use a file you provide, or draw a labelled box. Downloaded logos are cleaned of scripts before use.
    • A planned layout. One clear main flow, control above, operations below, nested boundaries, on an 8 px grid.
    • Automatic checks. Overlaps, lines through captions, labels on borders, crossing lines and text that doesn't fit are caught before you see the diagram.
    • The agent looks at it. It renders a PNG and reviews it before handing the diagram over.

    You get an editable .drawio file plus an SVG or PNG. The layout spec is kept next to them, so later changes are quick.

    Example prompts

    • "Draw the architecture of this repo for the README, with real Azure icons."
    • "Read the Terraform in infra/ and draw what it deploys."
    • "Make a reference architecture: API Gateway → Lambda → DynamoDB, with EventBridge fan-out to SQS and Step Functions."

    Free and open source: https://github.com/manaf-dev/architect-diagrams

    Needs Python 3.10+ and Chrome, Edge or Chromium to render previews.
    Needs internet the first time each icon is used.
    AWS and Kubernetes icons are native draw.io shapes: they show in the .drawio file and the PNG, but the standalone SVG shows labelled placeholders for them. Use the PNG for those diagrams.
    Rendering those icons to PNG uses draw.io's online viewer unless draw.io Desktop is installed.
    Layout is planned by the agent on a grid, not auto-laid out; very large systems are best split into several diagrams.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    • Free with an Agensi account

    Requires

    GitHub
    Jsdelivr
    Diagrams
    Python 3.10+
    Chrome, Edge or Chromium
    Python

    Permissions

    Terminal / Shell
    Write Files
    Browser
    Network Access
    Read Files
    Environment Variables

    Allowed Hosts

    raw.githubusercontent.com
    cdn.jsdelivr.net
    api.github.com
    viewer.diagrams.net

    File Scopes

    docs/architecture/**
    ~/.cache/archdiagram/**

    Runs Python scripts that build the diagram and write the spec, .drawio and image files (default docs/architecture/, or wherever you choose). Downloads icons from draw.io's GitHub, jsDelivr (Simple Icons) and the Fabric icon pack, cached in ~/.cache/archdiagram. Launches headless Chrome/Edge to render PNGs. Reads CHROME_PATH/DRAWIO_PATH only if set. No accounts, API keys or credentials needed. Every host and its purpose is listed in the "Network access and local programs" section of SKILL.md.

    Listed2 days ago

    What's inside

    Frequently Asked Questions

    Mermaid and most AI diagrams use generic shapes and automatic layout. This skill uses the real vendor icons, plans the layout on a grid, lints it for overlaps and crossing lines, and has the agent review a rendered image before delivering.

    Yes. You get a .drawio file that opens in draw.io or diagrams.net, plus the layout spec, so you can edit by hand or ask the agent to change the spec and re-render.

    AWS (952), Google Cloud (363) and Kubernetes (78) catalogs, draw.io's Azure, IBM and SAP SVG sets, the Microsoft Fabric community icon pack and 3,000+ brand logos from Simple Icons. You can also use your own SVG or PNG logos.

    An agent that can run Python (Claude Code, Codex, Cursor, Copilot, Gemini CLI and similar), Python 3.10+, and Chrome, Edge or Chromium for previews. PyYAML is optional, and internet access is needed the first time each icon is used.

    Yes. When you work in a repo, the agent reads your Terraform, Bicep, CloudFormation, CDK, Helm or docker-compose files to find the components and connections, instead of guessing. There is no automatic converter; the agent models the system from what it reads.

    The skill instructions (SKILL.md), three Python scripts (build and lint, icon lookup, render), design rules and spec reference, four example diagrams (AWS, Azure, Kubernetes, Microsoft Fabric), the icon catalogs, the MIT licence and third-party notices. It's free and open source.

    New in DevOps & Cloud