Brian Gershon

Building OpenClaw-style agents? Here is how to run skills without API keys in the agent container.

Pattern: two containers - agent posts to an internal URL, tool service holds the credential. The key never crosses into agent space.

www.briangershon.com/blog/securin...

Supercharging Claude Code with the Right (CLI) Tools Ten CLI tools recommended by Claude Code itself, ranked by how much they’d boost its productivity.

Claude Code was asked what tools would make it most effective.

Top picks: ast-grep for structural code search, difftastic for AST-based diffs, shellcheck for shell script safety. Tools that let AI work on code structure - not just text patterns.

batsov.com/articles/202...

Make the Tool You Wish Existed (with Your LLM) | Brian Gershon Build custom HTML tools that fit your exact workflow using LLMs. No extensive programming knowledge required. Turn workflow problems into practical tools in hours.

Thanks @simonwillison.net, always enjoy your content. I also really like HTML tools. Thank you for all the resources!

Was inspired to share them:

www.briangershon.com/blog/make-to...

Our first SeattleJS meetup of 2026 is coming up next Wednesday!

New year, new content format: We're going to be hosting a moderated panel discussion on using AI in developer workflows.

Curious about how other people are using LLM-based tools? Bring your questions for our panelists! RSVP:

Make the Tool You Wish Existed (with Your LLM) | Brian Gershon Build custom HTML tools that fit your exact workflow using LLMs. No extensive programming knowledge required. Turn workflow problems into practical tools in hours.

Want to try vibe coding without production software complexity?

HTML tools are the perfect playground for LLM-assisted development.

Single files, no deployment, no dependencies. Just experiment and iterate.

www.briangershon.com/blog/make-to...

Thanks @boscolo.co - I wish I had come up with “coding with no pants”, that’s an awesome title!

Securing AI Coding Tools: Permission Controls and Credential Protection for Engineering Teams | Brian Gershon AI coding tools handle security, data retention, and training data differently. This guide helps you quickly find tool-specific security configurations, compare approaches across Claude Code, GitHub C...

AI coding assistants index your entire codebase, including credentials and customer data.

Each AI coding tool handles security differently.

Configure exclusions and teach AI tools what's off-limits.

www.briangershon.com/blog/securin...

#AICoding #DevTools

Build a Registry of AI Tools. Generate MCP Servers and CLIs. | Brian Gershon Define prompts and tools in markdown files by describing them to an LLM. A registry framework converts those markdown specs into MCP servers and CLI tools automatically. You iterate in markdown, not f...

I wanted portable AI tools you can spin up by describing them, not hand-coding MCP servers.

Describe what you want once. A registry turns that into MCP servers and CLIs that any AI coding assistant can use.

One definition, every assistant and format you need.

www.briangershon.com/blog/build-a...

How to Correct AI Errors When Analyzing Customer Reviews | Brian Gershon When AI generates customer review summaries and trends, it hallucinates numbers and names. Learn how verification tools and feedback loops catch these errors automatically and teach AI to correct its ...

When AI summarizes customer reviews, it often invents names and numbers that look real but aren’t.

We saw this in weekly reports across Google, DoorDash, Yelp. Strong insights, wrong facts.

Fix: teach AI to verify its own work, not just summarize data.

www.briangershon.com/blog/feedbac...

GitHub - justincastilla/kitsap_commute_MCP: learning MCP through creating a commute assistant in Western Washington learning MCP through creating a commute assistant in Western Washington - justincastilla/kitsap_commute_MCP

Thanks @justincastilla.bsky.social, your MCP server puts the fun back into local travel. I plan to use it on my next trip to Gorst. github.com/justincastil...

Check Commands: Teaching AI to Catch and Correct Team Code | Brian Gershon When teams use AI tools to generate code, consistency becomes critical. Learn how the check command lets AI tools discover and fix their own errors, and how AGENTS.md teaches your AI tools to generate...

As teams adopt AI coding tools, they face familiar challenges: maintaining code quality and consistency with team standards. The same principles that help developers write better code can also guide AI tools.

Read more at: www.briangershon.com/blog/check-c...

#AI #DevTools

The second wave of MCP: Building for LLMs, not developers - Vercel The second wave of MCP, building for LLMs, not developers. Explore the evolution of MCP as it shifts from developer-focused tools to LLM-native integrations. Discover the future of AI connectivity.

Billions of requests across MCP servers deployed on Vercel, including Zapier, Solana, and our own.

A key insight: wrapping your API in an MCP server isn't enough.

Here's what's working best so far:
vercel.com/blog/the-se...

Donate to Honor Mikeal Rogers by Supporting His Family, organized by Anna Maier A Note from Mikeal’s friends… A Loving Father's Legacy Lives On We are… Anna Maier needs your support for Honor Mikeal Rogers by Supporting His Family

In support of Mikeal’s family, please donate to their GoFundMe if you can: gofund.me/0b3a1d72

I met Mikeal at the first NodeConf, which he helped organize. He’s the earliest and most vibrant memory I have of the excitement and community he helped form during the early days of Node.js, and as a strong voice in the JavaScript ecosystem. Rest in peace Mikeal.

A brief history of JavaScript | Deno In 30 years, JavaScript went from being a little scripting language to one of the world's most popular. Here are key moments to show how it has evolved and where it is headed.

I jumped full-time on the JavaScript train 2008, as Node.js was just starting up as well. Fun to run through this timeline and remember the birth of npm, Backbone (and spine.js), Atom Shell (Electron), React, Gulp, ...

deno.com/blog/history...

A handy tool for discovering and cleaning up code debt in your TypeScript projects:

`npx ts-unused-exports tsconfig.json`

Dos and don’ts when sunsetting open source projects Three maintainers share their tips for gracefully sunsetting open source projects.

How do you gracefully sunset an open source project? 🌇

Get valuable do's & don'ts from three experienced maintainers and set up a smooth transition for your community. 👇

Hey Doug, I’m curious about sampling.

Which MCP clients support this?

(My understanding is Claude Desktop doesn’t yet support)

One less rite of passage in our future!

Troubles with multipart form data and fetch in Node.js This is one of those cathartic blog posts. One in which I spent several frustrating hours trying to debug something that really should have just worked. Once...

If you are trying to use OpenAI Files API in Node.js and running into duplex and other errors uploading files:

You will fall down a lot of rabbit holes on the internet and with LLMs.

Instead, `npm add undici` to update Node to the latest version of `fetch`.

philna.sh/blog/2025/01...

Digital Space. How it should be Done We live in the digital epoch, but we keep our data on centralized services, with vendor lock-ins and constant policy changes, where your…

What are the risks and hidden costs of modern social networks and platforms? medium.com/@sergeyshand... #social-network #digital-rights #dweb #web3 #fediverse

Screenshot showing the "specifickeyword" filter in Firesky with a link to a RSS feed on the right

Bluesky / #atproto hack:

Filter posts about a rarely used topic (e.g. a small city you live in or any other *very* specific keyword) with Firesky like this:

firesky.tv?filter=speci...

Then click on the "(rss)" link, put this into your RSS reader.

Now you will get updates for this topic there.

Building software on top of Large Language Models I presented a three hour workshop at PyCon US yesterday titled Building software on top of Large Language Models. The goal of the workshop was to give participants everything they …

Here's the full workshop handout plus annotated slides from "Building software on top of Large Language Models", a three hour tutorial I presented yesterday at PyCon US #PyConUS simonwillison.net/2025/May/15/...

Congrats to @ngerakines.me for a new version of @smokesignal.events.

This is a very major rewrite, and also open sourcing of the whole codebase - with more features yet to come!

Events connect people - the exact kind of thing that should be available on an open protocol.

this might finally get me to publish sh.tangled.*

modelname.ai - AI Model Names Directory Find and copy AI model names instantly. Search OpenAI, Gemini, Claude & Grok models without digging through documentation.

A handy site for looking up the latest LLM models.

(though I didn't see the OpenAI o3 family in there)

modelname.ai

Product Purgatory: When they love it but still don’t buy When even “free” is too expensive.

Businesses intentionally allocate budget (includes both human beings and direct spend) on specific projects / initiatives / requirements / oppys / problems / etc..

They 𝘸𝘪𝘭𝘭 spend on whatever they've decided are the top few priorities.

(1/5)

ATmosphere Report – #114 Bluesky launches a checkmark verification system, streaming software Streamplace gets 500k USD in funding, and much more!

This week's Bluesky and ATProto news - ATmosphere Report #114

- First EU atproto conf with @ahoy.eu
- Verification on Bluesky with checkmarks
- @stream.place gets 500k in funding
- A 50k grant from @freeourfeeds.com for developing IndieSky with dev community

fediversereport.com/atmosphere-r...

Nice write-up!

“While chat is powerful, for most products chatting with the underlying LLM should be more of a debug interface – a fallback mode – and not the primary UX.

So, how is AI making our software more useful, if not via chat? Let’s do a tour.”