Founder of WolfMark Ltd
Independent software builder · AI & developer tools · automation · creator tech
I like turning “surely this can work better?” into software.
I'm an independent UK developer building products under WolfMark.
I work across developer tooling, AI-assisted workflows, automation, creator technology, desktop apps and the occasional idea that doesn't fit neatly into a category.
A lot of what I build starts the same way:
I run into something awkward, repetitive or unnecessarily complicated and wonder whether I can make it better.
I'm particularly interested in the point where AI agents stop being chatbots and start becoming genuinely useful tools — especially when they can work reliably with real files, code, workflows and other software.
I care about keeping things practical, understandable and useful rather than adding complexity just because we can.
Agent judgement for structure. Deterministic tooling for proof.
WolfMarkDown is an open-source Agent Skill for turning messy AI-generated Markdown into clean, professional documents that are ready to review and keep.
It can recover things AI agents commonly lose or damage, including:
- heading hierarchy
- comparison tables
- lists and labelled information
- conversational scaffolding
- code formatting
- technical identifiers
The agent handles the semantic judgement.
WolfMarkDown then provides deterministic verification using tooling including Prettier, markdownlint and GFM parsing, alongside integrity and idempotence checks.
Install:
npx skills add WolfMarkTools/WolfMarkDown --skill wolfmarkdown
WolfMark projects tend to sit somewhere around:
- 🤖 AI & agent tooling
- 🧰 Developer tools
- ⚙️ Automation
- 🎮 Creator & gaming tools
- 🖥️ Desktop software
- 📱 Apps
- 🧪 Experimental products and ideas
Some become products.
Some become open-source tools.
Some teach me enough to make the next thing better.
I tend to favour:
Small, useful products over giant platforms
Evidence over assumptions
Simple architecture over clever architecture
Automation over repetitive manual work
AI judgement where judgement is useful
Deterministic checks where correctness can actually be proved
And perhaps most importantly:
Ship it, use it, find out where it breaks, then make it better.
Right now I'm particularly interested in:
- Agent Skills and cross-agent workflows
- reliable AI-generated artefacts
- developer productivity
- local-first and desktop tooling
- AI-assisted software development
- practical automation
- tools that connect existing systems rather than replacing everything
WolfMark is my independent software and technology company.
The long-term goal is simple: build useful things that earn the right to keep existing.