Cha Khan · From the talk “It Just Keeps Getting Faster”, 19 Oct 2026

AI Resources

What I use every day to build with AI agents, in the order a piece of work goes.

Start hereELI5Any topic as a picture explainer. Install it first, then let it explain the rest of this list to you.Open →

Talk to your agent
Wispr Flow Enkeon
Agree on what to build
Grill Me How to be more ambitious HTML, not Markdown
Build it right
pstack HTML, not Markdown
Make it look good
taste-skill Impeccable diagram-design Anatomy
See the work
HTML, not Markdown ELI5Start here Lavish i-have-adhd AXI agent-browser
Ship it safely
no-mistakes
Run many agents
firstmate T3 Code herdr

What I find most useful

Voice to textI talk to my agent instead of typing. I say it messy and the agent sorts it out. It is faster, and I give much more context.

HTML, not MarkdownI ask for HTML pages instead of walls of text. I can see a plan, a diagram or a review at a glance, and click through it.

Or hand it to your agentIt installs ELI5, looks at the work you do, and makes you an ELI5 page on the tools that fit it.

Read resources.md. Install ELI5 from it right away. Then look at how I have been working and the work I have been doing, and pick the resources that would be most useful and relevant to me. Use /eli5 to make one ELI5 page that walks me through them: what each one does for me, why it fits my work, and the first thing to try. Ask before you install anything else.

Talk to your agent

Wispr Flow

Talk to your agent instead of typing. Messy is fine.

Wispr

Click or press → to step

Talk to your agent

Enkeon

One memory for you and all your AI tools. Context carries across sessions.

Enkeon

Open ↗

Agree on what to build

Grill Me

The agent interviews you until it understands what you want.

Matt Pocock’s skills

Click or press → to step

Agree on what to build

How to be more ambitious

Models can do more than most people ask. Aim higher, with specific goals.

Felix Rieseberg

speeddo the same work faster (the article skips this one)
altitudeaim at the outcome: what should happen next?
depthuse the real tools and raw data of top professionals
rangeborrow how other fields would solve it

Build it right

pstack

Makes the agent work like a careful senior engineer, and prove it.

Lauren Tan (poteto), Cursor

Click or press → to step

/poteto-mode picks a playbook for your request

Many agents working overnight, with a decision log

/how and /why answer with cited evidence

Make it look good

taste-skill

Art direction and reference images instead of generic AI pages.

Leon Lin

Open ↗

Make it look good

Impeccable

Critiques and polishes any interface: layout, type, colour, motion.

Paul Bakaus

Commands like /polish, /distill and /clarify

It spots and removes the usual AI design tells

Make it look good

diagram-design

Plain words in, a clean diagram out. 44 diagram types.

Cathryn Lavery

Make it look good

Anatomy

Explains an idea by drawing it as a real, working machine. Detailed isometric figures that move.

wheresryan22

Open ↗

Try/anatomy explain how a token bucket rate limiter works, interactive

Install

Claude Code

npx skills add wheresryan22/anatomy -g -a claude-code

Codex

npx skills add wheresryan22/anatomy -g -a codex

Agree on what to build · Build it right · See the work

The Unreasonable Effectiveness of HTML

Ask for an HTML page, not a wall of Markdown.

Thariq Shihipar, Claude Code team

Open ↗

Click or press → to step

See the workStart here

ELI5

Any topic as a picture explainer: big visuals, few words.

Thariq Shihipar, Claude Code community plugin

Open ↗

See the work

Lavish

An editor for your agent’s HTML pages. Click, change, send it back.

Kun Chen

Editing an agent’s HTML page in Lavish

See the work

i-have-adhd

Your agent leads with the answer and the next step.

ayghri

Before

Great question! Let me think about this. Your auth flow has a few moving pieces: the middleware, the token verification, and the cookie handling. Looking at src/auth.ts, the verifyToken function (around lines 42-58) seems to be using an older jsonwebtoken API. One approach would be to update the package and rewrite that function. After making the change, you'd want to run the auth tests to confirm nothing breaks. By the way, you might also want to look at your dependency versions overall. Hope this helps! Let me know if you want to dig deeper.

After

Edit src/auth.ts:42 to update the token validation. 1. Open src/auth.ts 2. Replace verifyToken (lines 42–58) with the snippet below 3. Run npm test -- auth.spec.ts Next: paste the first failing line if any test fails.

TryAsk anything. The answer comes first, steps numbered.

Install

Paste into your agent

Install the i-have-adhd skill/plugin from https://github.com/ayghri/i-have-adhd, refer to the repo's AGENTS.md for instructions.

See the work

AXI

Command-line tools built for agents: GitHub and Chrome, fewer tokens, fewer mistakes.

Kun Chen

Benchmark race: AXI CLI vs CLI vs MCP commands

Open ↗

See the work

agent-browser

Your agent opens, clicks and screenshots a real browser to check its work.

Vercel Labs

openagent-browser open localhost:3000
snapshotreads the page: buttons, links, fields
clickagent-browser click @e3
screenshotagent-browser screenshot page.png
checkthe agent looks, then fixes what is wrong

Frames from a recording, changes boxed in red

Ship it safely

no-mistakes

Make no mistakes.

Kun Chen

git push no-mistakes, start to finish

Open ↗

Run many agents

firstmate

Talk to one agent. It runs a crew and checks their work.

Kun Chen

You talk to firstmate. It runs a crew of agents.

yousay what you want, once
firstmateplans and splits the work
crewagents build the parts in parallel
checkevery part is reviewed
shipmerged, you get one report

Run many agents

T3 Code

One app for Claude, Codex, Cursor and more, side by side.

Theo, ping.gg

Threads on the left, agent in the middle, diff on the right

Open ↗

Run many agents

herdr

A terminal where many agents run in panes you can watch and talk to.

herdr