Mmcp.market

drawio-headless skill

by kevinnft·kevinnft/ai-agent-skills·14 stars·MIT

Generate architecture diagrams with draw.io in headless/server environments (WSL2, VPS, Docker)

A100/100content scan

Is the drawio-headless skill safe?

Clean: nothing in its files matched our rules. We read 4 files in the folder on 2026-09-28.

No findings.

Install the drawio-headless skill

A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description.

git clone --depth 1 https://github.com/kevinnft/ai-agent-skills.git /tmp/ai-agent-skills
mkdir -p ~/.claude/skills
cp -r /tmp/ai-agent-skills/skills/creative/drawio-headless ~/.claude/skills/drawio-headless
available in every project

In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub

The instructions your agent would load

SKILL.md as published, without the frontmatter. Read it on GitHub

Draw.io Headless Diagram Generation

Generate professional architecture diagrams from .drawio XML files in headless environments (WSL2, VPS, Docker) where GUI is not available.

When to Use

  • Server/VPS without X11 display
  • WSL2 Ubuntu (no native GUI)
  • Docker containers
  • CI/CD pipelines
  • Automated diagram generation

Prerequisites

Required packages:

# Install draw.io (snap - easiest)
sudo snap install drawio

# Install Xvfb (virtual framebuffer for headless rendering)
sudo apt-get update
sudo apt-get install -y xvfb

Alternative: .deb package

# Download latest .deb
wget https://github.com/jgraph/drawio-desktop/releases/download/v28.2.5/drawio-amd64-28.2.5.deb

# Install
sudo dpkg -i drawio-amd64-28.2.5.deb
sudo apt-get install -f  # Fix dependencies

Verification

# Check draw.io installation
which drawio
# Expected: /snap/bin/drawio or /usr/bin/drawio

# Test with xvfb-run
xvfb-run -a drawio --version
# Expected: draw.io version number (e.g., 28.2.5)

Usage

Basic Export

# Export .drawio to PNG
xvfb-run -a drawio -x -f png -o output.png input.drawio

# Export to SVG
xvfb-run -a drawio -x -f svg -o output.svg input.drawio

# Export to PDF
xvfb-run -a drawio -x -f pdf -o output.pdf input.drawio

Flags Explained

  • xvfb-run -a — Run in virtual framebuffer (headless)
  • -x — Export mode
  • -f — Output format (png, svg, pdf, jpg)
  • -o — Output file path
  • — Input .drawio file

Pitfalls

1. Snap Confinement (File Access)

Problem: Snap-installed draw.io cannot access /tmp or arbitrary directories due to confinement.

Solution: Use home directory or snap-accessible paths:

# ❌ FAILS (snap cannot access /tmp)
xvfb-run -a drawio -x -f png -o /tmp/diagram.png /tmp/input.drawio

# ✅ WORKS (home directory accessible)
cd ~/diagrams
xvfb-run -a drawio -x -f png -o diagram.png input.drawio

Snap-accessible paths:

  • ~/ (home directory)
  • /home//
  • /media/ (removable media)
  • /mnt/ (mounted filesystems)

2. OpenGL/dbus Warnings

Symptoms: Errors like:

libGL error: No matching fbConfigs or visuals found
dbus[]: Failed to connect to socket

Impact: These are warnings only — export still succeeds. Safe to ignore in headless environments.

Suppress (optional):

xvfb-run -a drawio -x -f png -o output.png input.drawio 2>/dev/null

3. Missing Xvfb

Symptom:

Error: Cannot open display: :99

Fix:

sudo apt-get install -y xvfb

4. Large Diagrams (Memory)

Problem: Complex diagrams with many elements may consume significant memory.

Solution: Monitor memory usage, increase if needed:

# Check available memory before export
free -h

# For very large diagrams, consider SVG (vector, smaller memory footprint)
xvfb-run -a drawio -x -f svg -o output.svg input.drawio

5. Premature Deletion (CRITICAL)

Problem: Deleting diagram files immediately after generation but before user confirms receipt.

Symptom: User reports "diagram not received" but files already deleted.

Root cause: Telegram/messaging platforms may have delivery lag. Deleting before send confirmation = data loss.

Solution:

# ❌ WRONG — Delete immediately after MEDIA: path returned
xvfb-run -a drawio -x -f png -o diagram.png diagram.drawio
echo "MEDIA:/path/to/diagram.png"
rm diagram.png diagram.drawio  # TOO EARLY!

# ✅ CORRECT — Keep files, cleanup weekly
xvfb-run -a drawio -x -f png -o ~/diagrams/analysis_$(date +%Y%m%d).png diagram.drawio
echo "MEDIA:~/diagrams/analysis_20260505.png"
# Files remain for user reference

# Weekly cleanup (manual or cron)
find ~/diagrams -name "*.png" -mtime +7 -delete

Policy: Never auto-delete diagrams after sending. User may need to reference them later or delivery may fail silently.

Diagram Types Supported

  • Architecture diagrams (microservices, cloud, infrastructure)
  • Flowcharts (process flows, decision trees)
  • Sequence diagrams (API calls, interactions)
  • Network diagrams (topology, connections)
  • ER diagrams (database schemas)
  • UML diagrams (class, component, deployment)

Example Workflow

1. Create .drawio XML

<mxfile host="app.diagrams.net">
  <diagram name="Example">
    <mxGraphModel dx="1200" dy="900">
      <root>
        <mxCell id="0"/>
        <mxCell id="1" parent="0"/>
        
        <!-- Add your diagram elements here -->
        <mxCell id="box1" value="Service A" 
                style="rounded=1;whiteSpace=wrap;html=1;fillColor=#dae8fc;" 
                vertex="1" parent="1">
          <mxGeometry x="100" y="100" width="120" height="60" as="geometry"/>
        </mxCell>
      </root>
    </mxGraphModel>
  </diagram>
</mxfile>

2. Export to PNG

xvfb-run -a drawio -x -f png -o diagram.png diagram.drawio

3. Verify Output

ls -lh diagram.png
file diagram.png
# Expected: PNG image data, ...

Integration with Hermes

When generating diagrams in Hermes workflows:

  1. Create .drawio XML programmatically (Python, script, template)
  2. Export via xvfb-run in execute_code or terminal tool
  3. Return path via MEDIA:/path/to/diagram.png for Telegram
  4. DO NOT auto-delete — Keep diagrams for user reference (cleanup weekly, not per-send)

Performance

Typical export times:

  • Simple diagram (5-10 elements): 2-3 seconds
  • Medium diagram (20-50 elements): 4-6 seconds
  • Complex diagram (100+ elements): 8-12 seconds

More skills from kevinnft/ai-agent-skills

  • Aaddyosmani-tddDrives development with tests. Use when implementing any logic, fixing any bug, or changing any behavior. Use when you need to prove that code works, when a bug report arrives, or when you're about to modify existing functionality.
  • AairtableAirtable REST API via curl. Records CRUD, filters, upserts.
  • Aapi-and-interface-designGuides stable API and interface design. Use when designing APIs, module boundaries, or any public interface. Use when creating REST or GraphQL endpoints, defining type contracts between modules, or establishing boundaries between frontend and backend.
  • Aapi-monitoring-botsBuild monitoring bots that poll APIs and send notifications on state changes (new listings, price alerts, status updates)
  • Aapple-notesManage Apple Notes via memo CLI: create, search, edit.
  • Aapple-remindersApple Reminders via remindctl: add, list, complete.
  • Aarchitecture-diagramDark-themed SVG architecture/cloud/infra diagrams as HTML.
  • AarxivSearch arXiv papers by keyword, author, category, or ID.
  • Aascii-artASCII art: pyfiglet, cowsay, boxes, image-to-ascii.
  • Aascii-videoASCII video: convert video/audio to colored ASCII MP4/GIF.
  • AaudiocraftAudioCraft: MusicGen text-to-music, AudioGen text-to-sound.
  • CaxolotlAxolotl: YAML LLM fine-tuning (LoRA, DPO, GRPO).

All agent skills → · MCP servers