A portable agent skill that enables LLM-powered coding agents to produce publication-quality, editable draw.io diagrams from papers, codebases, screenshots, and visual references through an iterative screenshot-feedback loop.
225
No data
8
0
MIT
2026-07-02
It addresses common failures of LLM-generated draw.io XML (text overflow, wrong arrows, missing icons) by enforcing a repeatable workflow with local preview, screenshot review, and defect logging — plus bundled icon assets and validation scripts.
A portable agent skill for producing publication-style, editable diagrams.net / draw.io figures from papers, prompts, codebases, project context, screenshots, and one or more visual references.
npx skills add Will-hxw/drawio-diagram-builder-skill
Copy this prompt into Codex, Claude Code, or another local coding agent:
Install and test the drawio-diagram-builder skill from:
https://github.com/Will-hxw/drawio-diagram-builder-skill
After installing, run its smoke test and tell me the exact skill path.
After installation, ask the agent to run the bundled update check against its installed skill path:
python \scripts\check_skill_update.py
| Requirement | Why |
|---|---|
| Python 3 (3.7+) | Preview and validation scripts |
| Browser automation (Playwright MCP, Puppeteer, browser tools, etc.) | Screenshot feedback loop — the skill is evidence-driven |
| Internet access | Preview loads https://embed.diagrams.net/ |
Without browser automation the agent can still generate .drawio XML, but cannot visually verify the result. The iterative refinement loop is the skill's main value.
LLMs can write draw.io XML, but the first result is usually not right:
This skill gives the agent a repeatable workflow: synthesize the user's text and image inputs into a diagram brief → create editable XML → preview through a local URL (not a giant encoded one) → screenshot → self-review visible and semantic defects → fix → repeat → validate.
For reference-image replication, the skill now enforces a stricter protocol: the agent must write a visual spec, coordinate layout grid, asset ledger, and defect log before drawing. Final handoff must include a screenshot-reviewed defect log, not just valid XML.
For prompt, paper, codebase, or mixed-input diagrams, the skill now also requires a self-supervision protocol: classify each input as content, structure, style, layout, or asset evidence; define connector semantics before drawing arro
Portable agent skill for research-style editable draw.io diagrams and screenshot-driven refinement
Similar projects matched by category, topics, and programming language.
Harness Engineering is a methodology for improving coding agent outputs by carefully crafting the environment around them—providing curated context, tools, and executable constraints that encode an organization’s nonfunctional requirements and cumulative lessons.
A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller entirely on-device, generating simple stories at about 9.5 tokens per second.
mimic captures traffic from any iOS or web app and automatically generates a Python client library that lets you call the app's API like a regular library.