Jake Ruesink

twitter init - #aboutme Good at: Javascript, CSS Scared of: DevOps, Spiders Work: Small dev agency (Lambda Curry) & Budgets.money

git worktree is a great feature for people smarter than me

I worked at Twitter during the 2010s. The place was a censorship machine, all in the name of “trust and safety”. Elon buying the company and rebasing it on absolute freedom of speech was a massive win for everyone, on both sides of the political debate.

@luisrudge @DrizzleORM @ryanflorence Came here just for this comment 😂

@DrizzleORM I guess drizzle is coming to Remix 3 cc @ryanflorence

guess what we're shipping next

@BrooksLybrand Just 2

Maybe the last human skill isn't taste Maybe it's giving 2 f*cks

@steipete @RatulSarna oh interesting, so it might be less if I turn off web extras?

@_jakeCodes @RatulSarna The codex statistics require a web view for scraping so that triggers this. I hope I can work on both sides to eventually find a better solution.

CodexBar 🎚️ 0.18 is out: - New providers: Kilo, Ollama, OpenRouter - Codex historical pace + risk forecasting + backfill - merged-menu Overview tab - fewer Claude keychain prompt annoyances - lower CPU/energy use, faster JSONL scanning thx @RatulSarna 🙏 github.com/steipete/Codex…

@tlakomy Is this actually true? Surely not by now.

Your periodic reminder that an average engineer is yet to try Cursor/Claude Code and you live in a bubble

@thewhiteriderX probably quite a few things will help initial thoughts on this: good structure with types and tests, good observability with alerts and warnings, agents that can apply checks and balances

@_jakeCodes If engineers become stewards of agent ecosystems, the hard problem becomes governance rather than prompts. How do you see identity and authority working between agents in those networks?

The role of engineers may be shifting. Less about executing every task. More about stewarding systems that execute work. Designing: • agent identities • communication pathways • governance • memory systems I wrote about this idea: lambdacurry.dev/blog/cultivati…

One mindset shift I’m encouraging on our team: Don’t just complete the ticket Close the stewardship loop: 1. Solve the problem 2. Notice what caused friction 3. Improve the system so it doesn’t happen again That’s where engineering leverage comes from lambdacurry.dev/blog/the-top-s…

The top skill engineers should be developing right now: *Stewarding the system. Instead of asking: “How do I complete this ticket?” Ask: “How can I make the system better so this problem is easier next time?” That’s where compounding impact comes from lambdacurry.dev/blog/the-top-s…

In the AI era, engineers aren’t just writing code. They’re designing the **machines that produce the code**. The engineers who learn to steward those systems will have the biggest impact. lambdacurry.dev/blog/the-top-s…

The biggest impact engineers have today isn’t how many tickets they complete. It’s whether their work makes the **next engineer faster**. That’s system stewardship. lambdacurry.dev/blog/the-top-s…

I currently have a multi-agent system running locally on a Mac Mini in my house. Agents with roles. Shared RAG memory. GitHub + Slack integrations. Continuous improvement loops. It started as a Codegen replacement. It turned into something much bigger. lambdacurry.dev/blog/cultivati…

A useful loop for engineers using AI: AI proposes a solution You challenge complexity You iterate toward elegance Then the important step: Update the **system** so the better pattern happens automatically next time. lambdacurry.dev/blog/the-top-s…

One surprising thing I've found while building agent systems: Identity matters. Giving agents narrative roles (builder, advisor, designer) dramatically improves coherence. LLMs understand stories. Architecture can use that. lambdacurry.dev/blog/cultivati…

AI doesn’t replace engineering judgment. It **amplifies** whatever engineering culture you already have. If your system is messy → you’ll generate mess faster. If your system is thoughtful → you’ll scale great patterns faster. lambdacurry.dev/blog/the-top-s…

I’ve been experimenting with looped agents. Agents that don’t wait for prompts — they run on cycles: • reviewing activity • proposing improvements • building features • updating documentation Eventually they may prompt us. More here: lambdacurry.dev/blog/cultivati…

What happens when you stop thinking about AI as a single assistant… …and start thinking about an ecosystem of agents? Builder agents. Advisor agents. Design agents. Shared memory. A system that improves itself over time. lambdacurry.dev/blog/cultivati…

A pattern I’m seeing with AI coding tools: They generate exactly what you ask for. The real problem often isn’t bad code. It’s **bad system context.** Small changes to prompts, rules, or architecture guidance can massively improve output. lambdacurry.dev/blog/the-top-s…