Everything I have open-sourced
The packages, agent skills and research code I have published as open source, in one place. I update it when I publish something new.
Here is everything I have published as open source, in one place. I update this page when I publish something new. Everything is under the MIT license unless noted.
agentmarkup
agentmarkup makes a website easy for AI to read. We built it to get the Anima Felix website close to a perfect AI readability score, then published it. The code is on GitHub.
- Plugins for Vite, Astro, Next.js and Nuxt, plus a CLI for any static site, all on top of @agentmarkup/core. They generate llms.txt, JSON-LD, markdown copies of the pages and AI crawler rules when you build the site.
- @agentmarkup/audit checks a live URL the way AI crawlers see it.
- A plugin for ChatGPT, Codex and Claude Code that walks you through setting it up and checking the result.
- AgentMarkup Studio, built with WebMCP: an AI agent in the browser edits a draft of your site’s machine-readable files while you watch, and rules check that the files agree with each other. How I built it.
Agent skills
- decent-skills:
/council-review, where several models review the same work without seeing each other’s answers, plus/wawaand/init-repo-docs. More about it. - agentmarkup-readiness-gap, the skill that won the GTM Skillathon. It compares what AI crawlers see on your site and your competitors’ sites with what a browser sees.
- Anxiety Reset by Anima Felix: brief grounding for a worry loop, as wellness support, not therapy. The skill, README and assets are under CC BY 4.0, and the code and plugin files under MIT.
Experiments
- infinite-slop-engine turns YouTube chat prompts into AI-generated TV clips, with moderation that blocks anything it cannot check. It ran the Interdimensional AI slop TV for one day, and I am not maintaining it.
Research code
Here is the code behind my papers, so anyone can rerun the tests. The four Perceived AGI prototypes are archived on Zenodo so they can be cited.
- scn-llm-prototype: time, paper 1.
- entropy-prototype: entropy, paper 2.
- truth-llm-prototype: truth, paper 3.
- love-llm-prototype: love, paper 4.
- eco-cpo-dag: code that uses contradictions to make supply chains accountable, from the paper ECO/CPO-DAG.
