Skip to content
View marmbiz's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report marmbiz

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
marmbiz/README.md

Hi, I'm Martin 👋

Senior Digital Platform Manager - CMS, UX, content governance & AI workflows.

I come from managing digital platforms, not from a pure software-engineering track: content systems, editorial workflows, SEO and governance for regulated industries such as health insurance, banking and biotech. With coding agents, I turn those operating problems into public tools and studies. The most visible public project right now is humanizer-de, a German AI Text Humanizer and Claude/Codex skill with 100+ GitHub stars, a 72-pattern catalog for German AI-writing signals, selected deterministic checks and evidence-safe rewrites.

🌐 martin-moeller.biz · 💼 LinkedIn


🛠️ Open source tools

Small, opinionated tools I build and use myself.

🧹 humanizer-de (100+ stars) - German AI Text Humanizer for Claude Code and Codex: 72 German-specific AI-writing patterns, selected deterministic linting and a 5-pass evidence-safe editorial review. Grew out of blader/humanizer.

🧩 fusion-lite - Local multi-model panels for running Claude, Codex, Gemini, Grok and other CLI/API tools from one terminal workflow.

🧭 future-self (new) - A decision-coherence skill for Claude Code and Codex. One question – "Will my future self thank me for this?" – and a five-stage exercise for when the answer dodges. Ends in kill criteria, not advice.


🔬 Research & platform audits

Projects on content operations, evidence and governance.

📊 What are Claude skills about? (new) - A metadata census of 8,277 non-fork Claude skill repositories on GitHub, classified by theme. Design leads at 28.5%, the median repo has 2 stars, and the field peaked in May 2026. No SKILL.md was read - metadata only, so there is no prompt-injection surface. Method, chart and raw data included.

🔍 AußenBlick GKV - An evidence-based audit of public health and benefit information across nearly the entire German statutory health insurance sector: 56,198 public pages, 84 web units. Language models prioritise pages for human review; they never produce the findings. Reproducible method, no ranking, no blame.

📄 Preprint: arXiv:2608.03500 (cs.CY) - LLM-Assisted Review Prioritization for German Statutory Health Insurance Websites: A Multi-Stage Corpus Audit. Open-access copy with DOI on Zenodo. A German short form is under submission at Monitor Versorgungsforschung.

🎤 Talk: "Wer verantwortet die Information, der Sie vertrauen?" - The study as a spoken walkthrough, in German.

✍️ The Lab - Notes on SEO, GEO and AI-content governance from running real platforms.

Pinned Loading

  1. humanizer-de humanizer-de Public

    German AI Text Humanizer for Claude Code & Codex. Audits 72 German AI-writing patterns using deterministic linters and evidence-safe rewrites. No fact-bending, no bypassing tricks.

    Python 121 11

  2. fusion-lite fusion-lite Public

    Local fusion-style multi-model panels that let existing CLI and API subscriptions do the heavy lifting.

    Python 2 1