Jayden Carey

@RhysSullivan @productdevbook github.com/productdevbook…

@RhysSullivan unadapter by @productdevbook might be relevant?

what are the best pluggable storage implementations you've seen that let people bring their own storage to your code? figuring out the public sdk shape for Executor, the simple path of static sources loading is straight forward, but dynamic tool loading requires state

@bwarrn @stylesshDev Just a heads up the GitHub link in the docs heading is broken/repo isn't public github.com/mesa/depot Mesa looks very cool!

@stylesshDev yep, just dm'd you some info

a couple of beers later... who's willing to use a modern git collaboration platform? - same core features as GitHub - Native AI layer with automations and agents. - Native support for stacked PRs. - focused, minimal and really fast.

@mikker This seems kinda like what @agentmail is, even though their positioning is AI focused.

One of you should build a reeeaaaally good email service that's just the API. So IMAP but also something more fancy like Gmail's or Fastmail's JMAP. Make it super fast, super reliable, sorta cheap. Make it easy to build fun stuff without the hassle of running email servers.

@shivam_nerd What provider are you using? I've got a GLM pro plan but it constantly gets rate limited, and even when it does it doesn't compare to Kimi K2.5. I'm sure it's a good model but the speeds and connection issues have made it practically unusable for anything but background work.

I think I am probably the biggest user of the GLM models at Factory. GLM-5.1 is a major improvement - super fast and significantly better. It is amazing as a daily driver, and super cost-effective as a worker model in missions. Try it out and let us know what you think!

@badlogicgames I know there's a redact function in github.com/superagent-ai/… , but it really just running the input you give it through an LLM with a "redact this data" prompt 😅 So it's a bit silly unless you're running the redaction model locally, and even then it's non-deterministic.

someone sent me a link to a typescript library for pii/sensitive data redactions a while ago. i can't find it anymore. only found this. github.com/sam247/openred… anyone got any pointers?

@badlogicgames As you've suggested, the before/after examples you included in the changelogs seem like more than enough to feed to an agent and make this pretty painless.

sorry for the pain this causes/will cause. but there's a reason pi isn't 1.0 yet.

People of pi. I'm emerging from the refactoring mines to give you the first pi release with some new internals. If you have extensions, please direct your agent at the changelog and updated docs, so it can fix it. Most of you will not be affected.

@rcarmo @micLivs @badlogicgames Yea this is what I've been doing too

@micLivs @badlogicgames Or you can cheat and run SearXNG in a container :)

Today I'm introducing not 1, not 2, but 12 🫣 new pi extensions. agent web search is a mess I'm not going to try to clean up. every provider has a slightly different tool shape, they share some common parameters but each has its own thing going for it. so instead of forcing one schema to rule them all, I'm shipping a package per provider (11) with their full native parameters, plus a router that auto-detects your API key and exposes just the common ones (query, country, language, numResults). `pi install pi-websearch-router` set the relevant environment key and it just works: Parallel, Brave, Exa, You.com, Tavily, Firecrawl, Jina, Linkup, Valyu, Perplexity, Serper, SerpAPI want provider-specific features like Exa's date filtering or Brave's freshness? install that provider directly instead. ``` pi install pi-websearch-parallel pi install pi-websearch-brave pi install pi-websearch-exa pi install pi-websearch-you pi install pi-websearch-tavily pi install pi-websearch-firecrawl pi install pi-websearch-jina pi install pi-websearch-linkup pi install pi-websearch-valyu pi install pi-websearch-perplexity pi install pi-websearch-serper pi install pi-websearch-serpapi ``` github.com/Michaelliv/pi-… @badlogicgames

@badlogicgames But still benefit of the doubt?

done did it. if you had any kind of interaction with the openclaw repo before (anything really), know that your issues/prs on pi-mono will be deprioritized and assumed bad by default.

@steveruizok You may not like it, but this is what peak performance looks like

cruising around 16K shapes at 120fps

@jachands @dillon_mulroy

I feel like I'm probably using better-result wrong @dillon_mulroy

@productdevbook Interested 👀

Built an API framework 🚀 Looking for 25 devs for closed beta — DM for access. —— is a type-safe, compiled-pipeline framework for next-gen TypeScript APIs.

I'm doubling down on state machines

@AonSayyed @ZixuanLi_ @Zai_org Yep deteriorates into gibberish rapidly past there. Hope that's fixed in 5.1

@ZixuanLi_ @Zai_org Meanwhile 5 is broken beyond 80k context

Don't panic. GLM-5.1 will be open source.

@zeeg Also xstate

@zeeg ts-pattern, permix, better-auth, Chat SDK

I need some great examples of idiomatic code in languages like TypeScript e.g. orpc/trpc-like interfaces where its very easy to reason about, low boilerplate, "obvious" on the surface and fairly explicit in behavior What are your favorite?

@dazuoforyou @zeeg Turn your slop reply stealer off dude x.com/i/status/20339…

@zeeg the abort controller discipline in pi is genuinely impressive. most agent frameworks just let operations hang forever. proper cancellation is the difference between a demo and a production tool. been diving into pi's codebase this week — the extension system is equally clean.

@aiwithjainam Great original post x.com/sentientt_medi…

10 MCP servers worth installing right now: 1. filesystem — read/write local files 2. github — full repo operations 3. postgres — query your database 4. brave-search — live web search 5. puppeteer — browser automation 6. slack — send messages programmatically 7. notion — read/write your workspace 8. linear — manage issues + projects 9. memory — persistent key-value store 10. sequential-thinking — structured reasoning All official. All free. All one command to install. MCP is the USB-C of AI tools. Everything plugs in the same way.

Eh. This is what I had been doing until recently, but recently made the switch to having a dedicated personal one. As a part of onboarding for my last role, for one reason or an other I was required I needed to change my primary email to be the work one for which was annoying. I realised could change it back later without breaking anything, but was still a requirement at the time which. There was one particular quirk I encountered though which triggered me to create a separate dedicated personal account. My org without my input decided to sign me up for a lower tier GH Copilot subscription, which automatically caused my existing personal subscription to be cancelled against my will (which at the time I'd purchased for a heavily discounted price), and prevented me from being able to resubscribe as well. No idea if that's been fixed at all now, but in that sense it stops feeling like it's still "your" account. I guess the main premise of this article is wanting to keep all the green squares on your contribution graph, which ultimately I don't really care about, and if I ever thought it did matter, they can very easily be faked 😅

Keep one GitHub account for your entire career. Your company's "create a new account" policy is wrong. That account is yours. You can rename it, change the password, add any email. Pro tip: star every private repo you contribute to. When you lose org access, starring keeps your green squares. georgeguimaraes.com/one-github-acc…

@PingStruggles Thoughts on adding a checkbox to show which files have been marked as viewed/reviewed? I quite like this feature of the GitHub VSCode extension. Happy to make a PR if interested.

Improved the PR diff sidebar UI in Better Hub 🙃