// resources

Everything I share

Links to the skills, tools and setup I use. Click through to the source.

My videos and everything on this page are my own opinions and don’t reflect my employer.

// skills

Skills

Skill collections I read and pull from. A skill is a folder with a SKILL.md that teaches an AI agent how to do one job.

Anthropic’s example skills plus a template. A good place to see how a skill is put together. The docx, pdf, pptx and xlsx skills are source-available, not open source.

Many skills Apache-2.0 (per its README)

A big, popular list of Claude skills and tools. Good for browsing; check each skill on its own, since each may have its own license.

Apache-2.0 (per its README)

How skills work in OMP, the coding agent I use, and how to write your own.

OMP is MIT

Where OMP skills get published. Browse by topic or see what people install most.

Set by each skill’s author

One engineer’s own day-to-day skills: planning, tests first, handoffs, writing.

MIT

A plugin in Cursor’s plugins repo full of engineering skills and principles. I’m pulling a lot from it, and I like its unslop skill, which cuts AI filler out of writing.

MIT (pstack’s own LICENSE file)

Listing a collection here doesn't mean I've vetted every skill inside it. When I download a skill, this is what I do:

  1. 1. SourceOpen the repo. Check the license, the last update and who maintains it. No license means you can link to it, not copy it.
  2. 2. Download and read itRead the whole SKILL.md and every file it ships. Look for network calls, shell commands, file writes outside the project, and anything that touches keys or passwords.
  3. 3. Scripts off by defaultAny plugin, hook or script stays turned off until I fully understand what it does and how I’d use it.
  4. 4. Test itTry it in an empty scratch folder with no real data. Run one task, watch what it touches, and keep it only if it did what it said and nothing more.
  5. 5. Make it mineCustomize the skill to my own structure and conventions, so it works better for how I actually work.

// prompts

Prompts

Reusable prompts I run every day.

Coming soon. I’m rewriting my prompts so anyone can use them.

// use-cases

Use cases

Real things I do with AI.

Plain notes as AI memory

#

I keep what my AI needs to know in plain markdown files. Any model can read them, so I can swap Claude for DeepSeek and keep my context.

Watch on YouTube:Part 1

Also on:TikTokFacebook

Local AI on a gaming PC

#

My gaming PC runs a Qwen model with Strata on an RTX 3080, so it does useful work when I’m not gaming.

Watch on YouTube:Part 1Part 2

Also on:TikTokFacebook

Talking to agents with dictation

#

I use local dictation on my Mac to talk to coding agents. Talking gives them more context than a short typed prompt.

Watch on YouTube:Part 1Part 2

Also on:TikTokFacebook

OMP’s usage charts put my work on the wrong day because of my timezone. I worked with an agent to find the cause, then wrote the fix and its tests on my fork.

Multi-agent workflows

#

One agent plans and splits the work; others build and review in parallel. More on this soon.

Watch on YouTube:Short

Also on:TikTokFacebook

// tools

Tools I use and recommend

What I actually run. Some links are referral links: you get a discount or a free trial, and I may get credit too.

My daily coding agent. Works with many models and supports skills and subagents.

MIT

Anthropic’s agents. I run Claude Code in the terminal and Cowork in the Claude desktop app.

Not open sourceReferral link

A guest pass: 7 days of Claude Pro free (Claude Code and Cowork included) if you’re new to paid Claude. Limited passes, so it may run out.

Try it:Claude CodeCowork

Where I view my markdown files. It’s the easiest way to see how my agents are structuring my notes.

Not open source

A fast, native terminal workspace from Stencil, the team behind OMP.

Closed beta (waitlist)

Runs a Qwen model on everyday hardware, with a one-click install and a local API.

MIT

An easy way to download and run open models on your own computer.

MIT

Hosted open-source models with an OpenAI-compatible API, so I can use them in my own tools.

Referral link

Sign up with my link and get up to $10 off your first month.

The framework this site is built with.

MIT

Utility-first CSS. It styles this site.

MIT

Hosts and deploys this site.

// vault

Vault structure

How my notes are set up so any AI agent can find its way around.

The starter kit (templates, scripts and starter skills) is coming soon for you to download and use.

  1. 01

    Conventions file

    One short file the AI reads first: folder layout, note fields, and where new info goes.

  2. 02

    Ingest

    Old notes and chats get sorted into the right hub instead of piling up.

  3. 03

    Hubs

    One note per area of life or work, with a short “current state” rewritten in place.

  4. 04

    Logs

    Dated lines under each hub, newest first, never edited. History stays, the summary stays short.

  5. 05

    Health check

    A script flags stale summaries, broken links and overdue tasks.

  6. 06

    Maintenance

    A daily start and a daily close keep hubs and plans current.

// github

GitHub

Repos I fork and work on in the open.

Where I work on fixes to OMP, so you can see how I change and test a real open-source project.

MIT (same as OMP)

// videos

Videos

Where I post. Same clips, pick your platform.

Shorts on the AI tools I use and what I build with them.

The same short clips, on TikTok.

The same short clips, on Facebook.