@elliscs.bsky.social
Reposted
Thanks to the newly relaunched @sveltesociety.dev website, I just saw that @syntax.fm did a deep dive into SvelteKit remote functions sveltesociety.dev/video/svelte...
SvelteKit has solved data loading! - Svelte Society
SvelteKit’s new remote functions and why they finally solve the long-standing pain of page-level data in Svelte. They cover queries, forms, batching, caching,...
sveltesociety.dev
November 20, 2025 at 10:15 PM
Thanks to the newly relaunched @sveltesociety.dev website, I just saw that @syntax.fm did a deep dive into SvelteKit remote functions sveltesociety.dev/video/svelte...
Shower thought, @svelte.dev + @anthropic.com Agent SDK running on a persistent server
November 20, 2025 at 9:31 PM
Shower thought, @svelte.dev + @anthropic.com Agent SDK running on a persistent server
hate when @developers.google.com gives me homework. great work though!
November 20, 2025 at 7:04 PM
hate when @developers.google.com gives me homework. great work though!
Building with AI agents in production, you hit the same wall every time: context rot
open.substack.com/pub/engineer...
open.substack.com/pub/engineer...
Orc Arch
Delegating Execution to Tools and Subagents
open.substack.com
November 20, 2025 at 3:15 PM
Building with AI agents in production, you hit the same wall every time: context rot
open.substack.com/pub/engineer...
open.substack.com/pub/engineer...
2026 is gonna be the year of tool calling
Gemini 3 and Grok Sherlock are already competing on it, and the SOTA models from China are as well. It maps well to the 12 factors as well
github.com/humanlayer/1...
Gemini 3 and Grok Sherlock are already competing on it, and the SOTA models from China are as well. It maps well to the 12 factors as well
github.com/humanlayer/1...
November 19, 2025 at 9:48 PM
2026 is gonna be the year of tool calling
Gemini 3 and Grok Sherlock are already competing on it, and the SOTA models from China are as well. It maps well to the 12 factors as well
github.com/humanlayer/1...
Gemini 3 and Grok Sherlock are already competing on it, and the SOTA models from China are as well. It maps well to the 12 factors as well
github.com/humanlayer/1...
www.youtube.com/watch?v=-FmV...
I really enjoyed this one
I really enjoyed this one
Dave Ricks, CEO of Eli Lilly, on GLP-1s and the business of pharma
Dave Ricks, the CEO of Eli Lilly, the world's most valuable pharmaceutical company, sits down with John and Patrick to discuss the complex business of drug development. Dave explains the true origin…
www.youtube.com
November 19, 2025 at 8:46 AM
www.youtube.com/watch?v=-FmV...
I really enjoyed this one
I really enjoyed this one
An interesting thought experiment
Take a workflow and do it manually. Find all the context needed to do it reliably.
Everywhere you get context is an opportunity for a subagent to only provide the same exact context you used to do it manually protecting the main agent
Take a workflow and do it manually. Find all the context needed to do it reliably.
Everywhere you get context is an opportunity for a subagent to only provide the same exact context you used to do it manually protecting the main agent
November 18, 2025 at 7:48 PM
An interesting thought experiment
Take a workflow and do it manually. Find all the context needed to do it reliably.
Everywhere you get context is an opportunity for a subagent to only provide the same exact context you used to do it manually protecting the main agent
Take a workflow and do it manually. Find all the context needed to do it reliably.
Everywhere you get context is an opportunity for a subagent to only provide the same exact context you used to do it manually protecting the main agent
the debate about AI creative writing seems a bit mid itself, i'm not looking for Hemingway in my email
November 18, 2025 at 3:30 PM
the debate about AI creative writing seems a bit mid itself, i'm not looking for Hemingway in my email
Reposted
More reliable way to make sure Claude Code skills are activated
How to Make Claude Code Skills Activate Reliably: scottspence.com/posts/how-to...
How to Make Claude Code Skills Activate Reliably: scottspence.com/posts/how-to...
How to Make Claude Code Skills Activate Reliably - Scott Spence
So, remember that
blog post I wrote
about Claude Code skills being a coin flip? Where the “simple hook”
approach gave me a whopping 50% succ...
scottspence.com
November 16, 2025 at 8:33 PM
More reliable way to make sure Claude Code skills are activated
How to Make Claude Code Skills Activate Reliably: scottspence.com/posts/how-to...
How to Make Claude Code Skills Activate Reliably: scottspence.com/posts/how-to...
These 2 updates from @svelte.dev are baller, time to update
createContext passes types with a stored context - replacing the need to type every getContext return value
The $state.eager(value) rune will update the UI immediately instead of waiting for the corresponding await to resolve
createContext passes types with a stored context - replacing the need to type every getContext return value
The $state.eager(value) rune will update the UI immediately instead of waiting for the corresponding await to resolve
November 16, 2025 at 3:14 PM
These 2 updates from @svelte.dev are baller, time to update
createContext passes types with a stored context - replacing the need to type every getContext return value
The $state.eager(value) rune will update the UI immediately instead of waiting for the corresponding await to resolve
createContext passes types with a stored context - replacing the need to type every getContext return value
The $state.eager(value) rune will update the UI immediately instead of waiting for the corresponding await to resolve
A pizza oven has been one of our best investments
November 16, 2025 at 1:49 PM
A pizza oven has been one of our best investments
Claude just shipped structured outputs in beta (Sonnet 4.5 + Opus 4.1). I kinda forgot they didn't have it because they pretty reliably respond correctly but if you need 100% accuracy then i guess you had to do the json parsing hacks
November 16, 2025 at 9:59 AM
Claude just shipped structured outputs in beta (Sonnet 4.5 + Opus 4.1). I kinda forgot they didn't have it because they pretty reliably respond correctly but if you need 100% accuracy then i guess you had to do the json parsing hacks
svelte.dev 's state management combined with factory pattern 🤩
Svelte
Web development for the rest of us
svelte.dev
November 15, 2025 at 4:18 PM
svelte.dev 's state management combined with factory pattern 🤩
Exploring the best abstraction for us to host the claude-agent-sdk. It is not straightforward. We particularly want persistent storage as I have a theory the SDK performs best with filesystem docs.claude.com/en/docs/agen...
Hosting the Agent SDK - Claude Docs
Deploy and host Claude Agent SDK in production environments
docs.claude.com
November 15, 2025 at 8:47 AM
Exploring the best abstraction for us to host the claude-agent-sdk. It is not straightforward. We particularly want persistent storage as I have a theory the SDK performs best with filesystem docs.claude.com/en/docs/agen...
Hints of shipmas coming around the corner!!
Gemini 3 spotted in the wild
Gemini 3 spotted in the wild
November 14, 2025 at 9:01 PM
Hints of shipmas coming around the corner!!
Gemini 3 spotted in the wild
Gemini 3 spotted in the wild
I noticed a pattern, agent frameworks are all racing to add governance layers now, we're doing it at work
LangChain just shipped human-in-the-loop middleware + tool-call limit controls. One demo literally shows "reining in a spend-happy shopping agent"
Runaway tool calls = runaway costs
LangChain just shipped human-in-the-loop middleware + tool-call limit controls. One demo literally shows "reining in a spend-happy shopping agent"
Runaway tool calls = runaway costs
November 14, 2025 at 7:33 PM
I noticed a pattern, agent frameworks are all racing to add governance layers now, we're doing it at work
LangChain just shipped human-in-the-loop middleware + tool-call limit controls. One demo literally shows "reining in a spend-happy shopping agent"
Runaway tool calls = runaway costs
LangChain just shipped human-in-the-loop middleware + tool-call limit controls. One demo literally shows "reining in a spend-happy shopping agent"
Runaway tool calls = runaway costs
thorny decoupling issues abound. we unwittingly coupled ourselves to Svelte4 with a pnpm workspace ui package of shadcn. it's been a beast to untangle and the only solution we've really found is pinning bits-ui internally in the package and side loading bits-ui directly in our app
November 13, 2025 at 9:31 PM
thorny decoupling issues abound. we unwittingly coupled ourselves to Svelte4 with a pnpm workspace ui package of shadcn. it's been a beast to untangle and the only solution we've really found is pinning bits-ui internally in the package and side loading bits-ui directly in our app
I haven't wrapped my head around models doing 200-300 sequential tool calls without human intervention like Kimi K2. I'm not sure I could do that so if this is SOTA we'll see a lot more of this 🧵
12 Factor Agents
github.com
November 13, 2025 at 7:04 PM
I haven't wrapped my head around models doing 200-300 sequential tool calls without human intervention like Kimi K2. I'm not sure I could do that so if this is SOTA we'll see a lot more of this 🧵
been using this a lot on my poorly written skills to great effect, thank you @scottspence.dev github.com/spences10/cl...
`Use pnpx claude-skills-cli to create a skill for my API client code
`Use pnpx claude-skills-cli to create a skill for my API client code
GitHub - spences10/claude-skills-cli
Contribute to spences10/claude-skills-cli development by creating an account on GitHub.
github.com
November 12, 2025 at 9:48 PM
been using this a lot on my poorly written skills to great effect, thank you @scottspence.dev github.com/spences10/cl...
`Use pnpx claude-skills-cli to create a skill for my API client code
`Use pnpx claude-skills-cli to create a skill for my API client code
We're starting to test the @anthropic.com Agent SDK as THE runtime
November 12, 2025 at 8:46 AM
We're starting to test the @anthropic.com Agent SDK as THE runtime
Debugging taxes kill iteration velocity.
Fast failure only works if learning is faster.
engineeredintelligence.substack.com/p/building-s...
Fast failure only works if learning is faster.
engineeredintelligence.substack.com/p/building-s...
engineeredintelligence.substack.com
November 11, 2025 at 3:31 PM
Debugging taxes kill iteration velocity.
Fast failure only works if learning is faster.
engineeredintelligence.substack.com/p/building-s...
Fast failure only works if learning is faster.
engineeredintelligence.substack.com/p/building-s...
Is it time to go whole hog on Claude skills and drop mcp? Feels too soon
November 6, 2025 at 9:20 PM
Is it time to go whole hog on Claude skills and drop mcp? Feels too soon
I posted this a little while back and it looks like Claire Vo just shared a rockstar masterclass on the same thing.
engineeredintelligence.substack.com/p/ephemeral-...
Vibe Analysis -> www.youtube.com/watch?v=KOr-...
engineeredintelligence.substack.com/p/ephemeral-...
Vibe Analysis -> www.youtube.com/watch?v=KOr-...
Ephemeral Software: UI, Data, and Functions in an AI-First World
What's "worth it" for Engineering to work on now
engineeredintelligence.substack.com
November 5, 2025 at 6:41 AM
I posted this a little while back and it looks like Claire Vo just shared a rockstar masterclass on the same thing.
engineeredintelligence.substack.com/p/ephemeral-...
Vibe Analysis -> www.youtube.com/watch?v=KOr-...
engineeredintelligence.substack.com/p/ephemeral-...
Vibe Analysis -> www.youtube.com/watch?v=KOr-...
Claude skills have changed nearly everything for me
November 4, 2025 at 7:57 PM
Claude skills have changed nearly everything for me