StarMorph

Agentic typescript engineering + software design
Claude | Codex | Hermes
practicing local inference

[StarMorph @StarmorphAI]

@beffjezos is this talk available to watch online?

[StarMorph @StarmorphAI]

@Pipe42uk @ssvankai 😂

[Alvaro Lozano-Robledo @mathandcobb]

Following up on the suggestion from Will Sawin, here is an illustration of the new configurations that disprove Erdos' unit distance conjecture (made with the help of ChatGPT 5.5 Thinking).

[Nick Frichette @Frichette_n]

@mattjay https://t.co/kzPMC47Qjw
Install this on your machine, all future install requests to NPM/PyPi will get routed through it. Malicious packages get blocked.

[André Bandarra @andreban]

WebMCP will enter the Origin Trial in Chrome 149, just a few weeks away. But you can try it today by enabling a Chrome flag!

Want to now more? Check out the docs at https://t.co/JkudOHKuqo

[Gagan Ghotra @gaganghotra_]

🚨 JUST IN - Google published a long piece about "Optimizing your website for generative AI features on Google Search" 👀

A lot in it https://t.co/22t75EtwUH

🧵

[Thomas Ricouard @Dimillian]

Our home in Codex mobile allows you to access any of your computers that are connected and logged on to the same account that your ChatGPT iOS app, but you can also connect by SSH to any other device running Codex!

[StarMorph @StarmorphAI]

@Trigun420 @JinjingLiang yes but what if you were using your entire claude sub for automated tasks? now limits are much lower

[StarMorph @StarmorphAI]

@imbabybrooklyn nous.dmg 🚀

[StarMorph @StarmorphAI]

@0xSero @TotallyNotParth visual explainer is 🔥
love agent browser too much smoother than vanilla playwright

[StarMorph @StarmorphAI]

@alex_whedon 👀 interesting

[StarMorph @StarmorphAI]

Just unlocked Diamond tier "Model Sommelier" in Hermes Agent ☤

@NousResearch · https://t.co/vJLCOZovgv

[Michael Guo @Michaelzsguo]

People are posting Qwen 3.6 configs that deliver fast TPS on as little as 12GB VRAM. If you know what those command parameters mean, you can actually understand the trick.

[StarMorph @StarmorphAI]

tip for using @NousResearch hermes - set auxiliary models for different tasks to optimize token usage

example provider: openai
---------------------
Set main model:
- model.provider = openai-codex
- model.default = gpt-5.5
- model.base_url = https://t.co/dblFnqXzMm

Set auxiliaries:
- compression = openai-codex / gpt-5.4-mini
- title_generation = openai-codex / gpt-5.4-mini
- session_search = openai-codex / gpt-5.4-mini
- skills_hub = openai-codex / gpt-5.4-mini
- approval = openai-codex / gpt-5.4-mini
- mcp = openai-codex / gpt-5.4-mini
- vision = openai-codex / gpt-5.4
- web_extract = openai-codex / gpt-5.4

change in .hermes/config.yaml

[StarMorph @StarmorphAI]

@_lopopolo following - very interesting would love to do this with my garden

[StarMorph @StarmorphAI]

@kyleanthony love your work

[StarMorph @StarmorphAI]

turboquant explanation
https://t.co/ArOJJF1pKx

[StarMorph @StarmorphAI]

@TheAhmadOsman I had a 3090 runnin Qwen in my cousins basement for 3 months. The thing got so hot it cured the gabagool. True story Tone hehe

[StarMorph @StarmorphAI]

@CardilloSamuel how was your experience comparing pi, opencode and hermes as a harness?