Dots Shmots. I've got @pidotdev durable working on a ChatGPT site with an MCP that allows agents to chat to it. Running forked Codex CLI with 4x less token consumption xD @badlogicgames (built 95% with Pi, dw)

People of @pidotdev! I present to you completely vibeslopped component library you might be able to just plug into your Pi durable. Will it work? I don't know. It's late. I haven't tried, its 1am here. Clankers say they work. Knock yourself out. More to come. Maybe. Happy Clanking!

recommended reading, though i still disagree, as explained on @syntaxfm early this year. for code, both memory and docs are not efficient. your codebase is all you need. make it modular, make modules small, at most, have a little maintained map of where what is.

I've slashed Codex's token startup cost 4 times while retaining full functionality. It works with ChatGPT desktop and is ready to tackle the quota abuse that the Dots will probably cause me. Works for me, will probably work for you. AMA. Compiled downloads and source code here : What cut the context overhead: - Stripped bloated system prompts and repeated tool instructions. - Replaced upfront tool inventories with compact usage hints and on-demand help. - Made skill discovery and loading on demand instead of injecting the whole catalog. - Load nested AGENTS.md by default instructions when reaching the relevant files. - Trimmed tool-output metadata and slimmed bundled skill instructions. Other highlights - /settings - you can switch some optimisations on and off from there. - all current experimental flags exposed in settings so you don't have to dig them up and then edit toml manually. - Persistent JavaScript/TypeScript Notebook with reusable variables and helpers across calls and context resets. Basically a persistent shell. - Notes-based OAI context management enabled by default for all models and several optimisations around that. - Thanks to this, all models can use their full 272k ctx. I made some precautions so they should hardly ever exceed it. - When you send a prompt after 25 mins, ctx window automatically rolls over before the prompt so your chat is never cold. Tried and tested, no risk to continuity at all. - Overall CTX management implementation vastly improved - I've been working on it for my Pi version since Astra released. - Agent board enabled by default. - Voice progress updates, context continuity (summarisation before engagement), and dropped-call recovery. - One personality file for text and voice (with a script adding voice personality to the desktop app - please note this changes every release so you'll have to vibe-update it), separate from AGENTS.md. OAI does not provide any of this. - Works as the backend for ChatGPT Desktop, locally or over SSH. Server works normally. You do not have to totally replace the Codex installation for this to work, the change to the ChatGPT desktop routing are 100% reversible. I suggest using .codex-lean as the global dir. Skills will need copying over as the skill tool supports categories. Instructions attached in the repo. Why do this? 1. Because I think Codex CLI could be more efficient 2. This is pretty much the best way to use Dots without messing around with herdr etc. 3. Yall like to whine about quotas. Yours truly has no issues with them, running basically this version of Codex, but in Pi.

3rd party personal assistants will soon all be F tier. Hear me out. I haven't yet used Dots. I haven't yet used Muse. I used Grokbot a bit. I've used Openclaw in gpt 5.4 era, was really meh. I've installed and uninstalled Hermes twice very quickly, I just did not jive with anything it was doing. I've clanked up my own persistent personal agent system that I am happy with, but I don't really have it in me to think too much about making it super fully fledged so it just runs Pi with some extras. With that out of the way, I think 3rd party assistants, like Hermes, OC or any other flavour of it (with ZERO preference for one over the other, this is not the point of this post), will soon all be F tier. Why? Think about it. Big companies will throw anything at their version to win the "bot wars", this includes both engineering power AND subsidies. I think this is obvious that unlimited tokens are very likely to win with an opensource project of any kind, as much as OSS has its advantages with "distributed engineering". Muse is apparently cheap as chips. Grok Bot has a separate quota from Grok Build and Cursor (at least for now). OAI are currently running Dots on a promo, although I'd say it's closer to early-access-beta-testing. Once they figure out how to subsidise these, we'll probably also see a separate quota or something. Third parties are left with people who: - are running models locally and want stuff super-private - are willing to just pay for cheap API - have some sort of a non-frontier subscription - are happy to spend their - to use this example, because it's already there - Grok sub quota on running a personal assistant they prefer over spending using the version of it that is drawing from a separate pot of quota allowance. - think that an always online, full internet access, personal assistant in a 3rd party harness is more private than something that runs on 1st party servers (here imma be judgemental - very naive of them) That leaves the 3rd party assistants in a very weird spot. Openclaw/Hermes had a head start. But Grok Bot has Poteto, etc etc. So they will soon all catch up and surpass the OGs and you'll have to be really hardcore to even consider still using something that's not 1st party. The gap will widen. Just rambling, but what I meant was: Dafuq you even posting these tier lists for? Just submit feedback/issue or write an actual review of each. OR MAYBE, OH GOSH, GET INVOLVED AND SUBMIT A PR WHERE YOU CAN?! No? Why? Cuz doing this doesn't farm engagement? I see.

Alright nerds, this is a must watch. Jorbster made a thing that plays and wins STS2 runs. As an addict with 2k hours in STS1&2 logged on steam and then some on the phone... :> The vid is a presentation+explanation+playthrough, don't be put off by the runtime