John Riccardi
@arcanegraindev.bsky.social
Lead Crafter at @8thLight
Mostly frontend, too many D&D side projects - TTRPG stuff here @arcanegrain.dev
https://arcanegrain.dev/
❤ Madison, WI ❤
Mostly frontend, too many D&D side projects - TTRPG stuff here @arcanegrain.dev
https://arcanegrain.dev/
❤ Madison, WI ❤
Pinned
John Riccardi
@arcanegraindev.bsky.social
· May 28
TDD: The Missing Protocol for Effective AI Assisted Software Development | 8th Light
Test Driven Development (TDD) provides a powerful framework for improving AI-assisted coding. Learn how writing tests first gives AI the clarity it needs to generate robust, reliable code—and boosts c...
8thlight.com
Decided to expand on some of my posts here and turn it into a blog post!
8thlight.com/insights/tdd...
8thlight.com/insights/tdd...
New blog post up on my experiment to set up Claude Code to help me mock out design ideas. Just part 1 as I learn, but I'll be posting more and hopefully post a full setup explanation once I get something stable.
arcanegrain.dev/blog/claude-...
arcanegrain.dev/blog/claude-...
Using Claude Code to Generate Design Mockups - Part 1: Goals & Setup
An ongoing experiment where I use Claude Code to generate design mockups based on text prompts.
arcanegrain.dev
January 14, 2026 at 11:15 PM
New blog post up on my experiment to set up Claude Code to help me mock out design ideas. Just part 1 as I learn, but I'll be posting more and hopefully post a full setup explanation once I get something stable.
arcanegrain.dev/blog/claude-...
arcanegrain.dev/blog/claude-...
Reposted by John Riccardi
Garbage Pile Friends
Making swag with that Flagrant funk
www.beflagrant.com
January 14, 2026 at 6:13 PM
Today's learnings so far
- Use the frontend-design plugin for Claude
- Use Playwright MCP server to allow Claude to "look" at it's output, creating a stronger feedback loop
- Create scripts to have claude take screenshots - don't waste tokens each time having it figure out how to take screenshots
- Use the frontend-design plugin for Claude
- Use Playwright MCP server to allow Claude to "look" at it's output, creating a stronger feedback loop
- Create scripts to have claude take screenshots - don't waste tokens each time having it figure out how to take screenshots
January 13, 2026 at 7:16 PM
Today's learnings so far
- Use the frontend-design plugin for Claude
- Use Playwright MCP server to allow Claude to "look" at it's output, creating a stronger feedback loop
- Create scripts to have claude take screenshots - don't waste tokens each time having it figure out how to take screenshots
- Use the frontend-design plugin for Claude
- Use Playwright MCP server to allow Claude to "look" at it's output, creating a stronger feedback loop
- Create scripts to have claude take screenshots - don't waste tokens each time having it figure out how to take screenshots
Welp, turns out the Claude skill I was creating already exists as a Claude plugin! I'm not exactly sure how plugins work vs skills and agents though. I think it's essentially just a skill packed up for you. Building a prototype with it now!
github.com/anthropics/c...
github.com/anthropics/c...
claude-code/plugins/frontend-design at main · anthropics/claude-code
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflo...
github.com
January 13, 2026 at 4:14 PM
Welp, turns out the Claude skill I was creating already exists as a Claude plugin! I'm not exactly sure how plugins work vs skills and agents though. I think it's essentially just a skill packed up for you. Building a prototype with it now!
github.com/anthropics/c...
github.com/anthropics/c...
Of course after posting a link to my site, I start having DNS issues. I think the autorenewal at Namecheap messed some things up.
January 10, 2026 at 4:24 PM
Of course after posting a link to my site, I start having DNS issues. I think the autorenewal at Namecheap messed some things up.
Took a little bit of tinkering (I think it was a pnpm issue), but I created a Claude "skill" to offload some info to. It's a visual inspection skill that will run the Playwright MCP server and do visual checks of my project, including checking breakpoints.
github.com/riccjohn/blo...
github.com/riccjohn/blo...
blog/.claude/skills/visual-inspection at main · riccjohn/blog
Contribute to riccjohn/blog development by creating an account on GitHub.
github.com
January 9, 2026 at 10:26 PM
Took a little bit of tinkering (I think it was a pnpm issue), but I created a Claude "skill" to offload some info to. It's a visual inspection skill that will run the Playwright MCP server and do visual checks of my project, including checking breakpoints.
github.com/riccjohn/blo...
github.com/riccjohn/blo...
So far I'm using an extremely simple setup with Claude. Just a CLAUDE.md and the Playwright MCP server. I told Claude (in the CLAUDE.md file) to make use of the MCP server and capture screenshots at mobile, tablet, desktop viewport sizes when it needs to verify that a new feature works.
January 9, 2026 at 4:46 PM
Reposted by John Riccardi
LLMs are not sentient, are not human. Grok cannot “apologize” or explain how or why any of this is happening. But of course major outlets such as Bari Weiss’s CBS News wrote that Grok “acknowledged ‘lapses in safeguards'." @jasonkoebler.bsky.social with more:
Listen: www.youtube.com/watch?v=zT9l...
Listen: www.youtube.com/watch?v=zT9l...
January 7, 2026 at 3:33 PM
LLMs are not sentient, are not human. Grok cannot “apologize” or explain how or why any of this is happening. But of course major outlets such as Bari Weiss’s CBS News wrote that Grok “acknowledged ‘lapses in safeguards'." @jasonkoebler.bsky.social with more:
Listen: www.youtube.com/watch?v=zT9l...
Listen: www.youtube.com/watch?v=zT9l...
I'm going to be diving into Claude Code over the next few weeks. I feel like I've been using these tools pretty naively and want to learn more about subagents, skills, etc. I'll be posting here as I learn and possibly get a small project set up and/or fix a bug on my site that I just found!
January 7, 2026 at 4:30 PM
I'm going to be diving into Claude Code over the next few weeks. I feel like I've been using these tools pretty naively and want to learn more about subagents, skills, etc. I'll be posting here as I learn and possibly get a small project set up and/or fix a bug on my site that I just found!
My new TV has AI built in for some reason, but it still can't make Tom Brady likeable in real time.
November 27, 2025 at 7:08 PM
My new TV has AI built in for some reason, but it still can't make Tom Brady likeable in real time.
Reposted by John Riccardi
i switched back to writing code by hand in my hobby project for an interesting reason... it's really psychological but when the AI gets stuck (e.g. the problem is too tricky to debug without good tools and atm it requires a human, or something got too messy), i get a mental freeze digging into code
November 23, 2025 at 3:57 PM
i switched back to writing code by hand in my hobby project for an interesting reason... it's really psychological but when the AI gets stuck (e.g. the problem is too tricky to debug without good tools and atm it requires a human, or something got too messy), i get a mental freeze digging into code
Reposted by John Riccardi
"It's really good at summarising code I don't understand".
But, weirdly, it's pretty bad at summarising code I *do* understand.
But, weirdly, it's pretty bad at summarising code I *do* understand.
November 24, 2025 at 7:33 AM
"It's really good at summarising code I don't understand".
But, weirdly, it's pretty bad at summarising code I *do* understand.
But, weirdly, it's pretty bad at summarising code I *do* understand.
Reposted by John Riccardi
Reposted by John Riccardi
"fearing or regulating it, or opposing technological progress, would hasten the coming of the Antichrist, Thiel said"
what the fuck
what the fuck
New argument against AI regulation just dropped. Checkmate, atheists.
www.wsj.com/tech/peter-t...
www.wsj.com/tech/peter-t...
September 24, 2025 at 2:04 PM
"fearing or regulating it, or opposing technological progress, would hasten the coming of the Antichrist, Thiel said"
what the fuck
what the fuck
Reposted by John Riccardi
Reposted by John Riccardi
QA by inspection is a mugs game, anyone who has been in operations will tell you this. It loses efficacy after about 30 minutes, your brain will "fill in" what it is expecting to see and miss the error.
March 28, 2025 at 5:27 PM
QA by inspection is a mugs game, anyone who has been in operations will tell you this. It loses efficacy after about 30 minutes, your brain will "fill in" what it is expecting to see and miss the error.
I'm starting to feel like the reliance on "you have to review all the AI generated code, of course!" as a way to mitigate the issues with AI is not really said in good faith. It's a CYA clause.
1/?
1/?
September 16, 2025 at 8:26 PM
I'm starting to feel like the reliance on "you have to review all the AI generated code, of course!" as a way to mitigate the issues with AI is not really said in good faith. It's a CYA clause.
1/?
1/?
Reposted by John Riccardi
Hi authors! Please remember to file a claim!
AI advocates have warned that if every author in the class action filed a claim, it would "financially ruin" the entire industry.
Authors celebrate “historic” settlement coming soon in Anthropic class action
Advocates fear such settlements will “financially ruin” the AI industry.
arstechnica.com
August 27, 2025 at 5:30 PM
Hi authors! Please remember to file a claim!
Reposted by John Riccardi
This is completely insane and a sign of how washed A.I. truly is. Why do you have to threaten people to use this if it's so good
CEO Brian Armstrong mandated that all Coinbase engineers set up their GitHub Copilot or Cursor accounts within a week.
Coinbase CEO explains why he fired engineers who didn’t try AI immediately | TechCrunch
CEO Brian Armstrong mandated that all Coinbase engineers set up their GitHub Copilot or Cursor accounts within a week.
techcrunch.com
August 24, 2025 at 12:47 PM
This is completely insane and a sign of how washed A.I. truly is. Why do you have to threaten people to use this if it's so good
zed.dev/blog/dialing...
A few thoughts about this post.
1. I completely agree with the take that we really need a realistic view of what these tools (that's right, they're TOOLS) can and more importantly, cannot do.
1/?
A few thoughts about this post.
1. I completely agree with the take that we really need a realistic view of what these tools (that's right, they're TOOLS) can and more importantly, cannot do.
1/?
Why I'm Dialing Back My LLM Usage - Zed Blog
From the Zed Blog: Alberto Fortin shares his honest reflection on the reality of using LLMs in production code and why he's taking a more measured approach.
zed.dev
July 9, 2025 at 8:53 PM
zed.dev/blog/dialing...
A few thoughts about this post.
1. I completely agree with the take that we really need a realistic view of what these tools (that's right, they're TOOLS) can and more importantly, cannot do.
1/?
A few thoughts about this post.
1. I completely agree with the take that we really need a realistic view of what these tools (that's right, they're TOOLS) can and more importantly, cannot do.
1/?
www.honest-broker.com/p/the-force-...
"The AI business model would collapse overnight if they needed consumer opt-in."
"The AI business model would collapse overnight if they needed consumer opt-in."
July 6, 2025 at 3:07 PM
www.honest-broker.com/p/the-force-...
"The AI business model would collapse overnight if they needed consumer opt-in."
"The AI business model would collapse overnight if they needed consumer opt-in."
Need a little index card sized character sheet for OSE? Here ya go! 🧙🏼♀️ arcanegrain.dev/downloads/
Arcane Grain
John Riccardi | Lead Software Engineer specializing in React, TypeScript, and frontend development.
arcanegrain.dev
July 6, 2025 at 3:09 AM
Need a little index card sized character sheet for OSE? Here ya go! 🧙🏼♀️ arcanegrain.dev/downloads/