Byron Delgado

[Byron Delgado @delgadobyron]

@OVO_Kareem @Itsfoss These options re build also for AI agents, they support MCP protocol

[Byron Delgado @delgadobyron]

@thsottiaux Provide similar capabilities to graphify

[Byron Delgado @delgadobyron]

@notkevinzhang Better Solution Architecture and estimations analysis

[Griff @ghagler]

"Opus 5 got worse" is the loudest complaint in AI right now. Mostly, it didn't — your setup did.

Anthropic stripped 80% of Claude Code's system prompt for this generation. The verification steps, double-check rules, and scaffolding teams spent a year writing now compound with behavior the model already does on its own. The fix for half the problems is deleting instructions, not adding them.

We put everything into one free GRIFFai field report for Claude users: the 20 complaints people are actually filing, sorted by what fixes each — instruction, setting, workflow, or nothing. Copy-paste blocks included, plus the delete list to run against your prompts first.

Built for practitioners. No fluff, 11 pages.

https://t.co/3hs8CIpSn8

#Claude #ClaudeCode #AIEngineering #GRIFFai

[Byron Delgado @delgadobyron]

@karpathy Is this be beggining of the end of SaaS?

[Peter Steinberger 🦞 @steipete]

How the sauce is made @openclaw

[OpenClaw🦞 @openclaw]

🦞 OpenClaw 2026.1.30

🐚 Shell completion
🆓 Kimi K2.5 + Kimi Coding: run your claw for free
🔐 MiniMax OAuth: one more model just a login away
📱 Telegram got a glow-up — 6 fixes from threading to HTML rendering

Plus a bunch of community-contributed fixes across LINE, BlueBubbles, routing, security & OAuth.

The lobster provides 😏 https://t.co/KyWuiuTzos

[Byron Delgado @delgadobyron]

@openrouter @openclaw @kilocode Any chance Claude Code?

[Byron Delgado @delgadobyron]

@bcherny Loving the Tasks feature. However, I had an issue: background tasks hit the compact threshold, and then everything became unresponsive. Claude was advising me to restore a previous checkpoint.

[Byron Delgado @delgadobyron]

@mattpocockuk Have you tried GSD?

https://t.co/eg2f5AP2hA

[Byron Delgado @delgadobyron]

@lennysan Anthropic shared this document with all its internal use cases for using Claude Code

https://t.co/8K2kIbmrim

[Byron Delgado @delgadobyron]

@windsurf_ai Gemini Flash 2.5, costing under 1 credit, please

[Byron Delgado @delgadobyron]

@kevinhou22 @jordankdalton @windsurf_ai Gemini please

[Byron Delgado @delgadobyron]

@MervinPraison @sama High expectations there! Not pressure 😅

[Chubby♨️ @kimmonismus]

In my opinion, @AnthropicAI is the biggest competition for @OpenAI. Interestingly, they also manage to attract senior OpenAI staff like Jan Leike. Anthropic has a certain magic and especially a strength that allows a four-month-old model to remain at the top, even though it is only a mid-tier model and outperforms almost all other LLMs (Sonnet 3.5).
Anthropic is currently valued at $40-50b, making it a serious competitor financially, especially as the company is backed by Amazon.
All in all, it will be very exciting to see what the upcoming Opus 3.5 will allow us to do, especially whether it is just a scaled model or whether it also features methods similar to CoT. In any case, they have gained the research staff for this.

[Byron Delgado @delgadobyron]

@theDushi1 @cursor_ai Simple, yet effective 👏

[Byron Delgado @delgadobyron]

https://t.co/mUayv0Leip

[Byron Delgado @delgadobyron]

https://t.co/NLXI1XKf7K

[Byron Delgado @delgadobyron]

https://t.co/udQQ2dQV5P

[Byron Delgado @delgadobyron]

https://t.co/PJYytiWGLT