Wade4AIWeb3

AI|Crypto| @hey_maybe_ai @Footprint_Data

[Wade4AIWeb3 @no7wade]

again n again...

[!HAN @hanyz_eth]

@OpenAIDevs @OpenAI I’m currently unable to use Codex because the selected model keeps returning: “Selected model is at capacity. Please try a different model.”

Is this a known capacity issue? Is there an ETA, or a recommended model to switch to? Screenshot attached. Thanks.

[Wade4AIWeb3 @no7wade]

@alliekmiller the PR accepted including those merged local (count not token wise) is probably the only one metric that i would accept as a developer

[Wade4AIWeb3 @no7wade]

@NoamShazeer A rough historical pattern for GOOGL after a panic selloff of 5%+ is:

[Wade4AIWeb3 @no7wade]

https://t.co/Mu1yWsM2K6

[Wade4AIWeb3 @no7wade]

https://t.co/XKAwlD7Iw0 go

[Vaibhav (VB) Srivastav @reach_vb]

This is now fixed along with the latest release of Codex!

Make sure to upgrade your codex installation to the latest version via npm or bash installer

Thanks again to all of you for raising this issue and to the goated (jif) codex team

[Wade4AIWeb3 @no7wade]

new stickers

[Wade4AIWeb3 @no7wade]

i was looking for data how long until a business becomes profitable?
Hermes Agent give me this following image(saying no single data for this):
(I'm in Info/Tech😂)
source https://t.co/86CHAKOkqN
@stats_feed u might love it

[Wade4AIWeb3 @no7wade]

A simple explanation:
Harness engineering means designing the system around an AI so it can reliably achieve goals, not just answer prompts. It includes tools, planning, memory/context, execution loops, verification, retries, and failure handling.
If prompt engineering is "make the model respond better," harness engineering is "make the whole AI work system finish the job(via multiple run/hop)."

[Wade4AIWeb3 @no7wade]

LLM Wiki vs RAG is not really “wiki replaces RAG.”
The sharper distinction is scale and governance:
- LLM Wiki works well for small, stable, curated knowledge
- RAG works better for large, dynamic, multi-user - enterprise corpora
- Neither solves bad source data, access control, freshness, or lineage alone

Best architecture: wiki for curated context, RAG for retrieval at scale.
Source:
https://t.co/p0KacQMIIk

[Wade4AIWeb3 @no7wade]

https://t.co/y3dYHwv9ki

[Wade4AIWeb3 @no7wade]

I used OpenAI Codex’s new Record & Replay-style workflow to turn a messy X Article publishing process into a reusable skill.
Took ~30 min end to end. ~6 runs conversation
Model: GPT-5.5 Codex. no script generated , just skill.md
here is Final skill: https://t.co/Y71M79QmAn
here is the article i published via this skill https://t.co/7miLY9f26X
Problems hit:
- X login/browser automation blockers (it switched to chrome dont know why, i use Brave browser )
- wrong draft focus
- body paste accidentally targeting title
- too many blank lines
- Markdown - item not becoming real X bullets( solved after at lease mentioned it twice )
- Preview button unreliable
Human helped by logging in, clicking the right draft, and spotting formatting issues with screenshots.
Codex fixed it by:
- collapsing extra newlines
- converting Markdown lists into native X lists
- removing leftover -
- using direct /preview
- updating the reusable skill

Took ~30 min end to end. ~6 runs conversation
Model: GPT-5.5 medium Codex.
I used OpenAI Codex’s new Record & Replay-style workflow to turn a messy X Article publishing process into a reusable skill.
here is Final skill: https://t.co/Y71M79QmAn
here is the article i published via this skill https://t.co/7miLY9f26X
Problems hit:
- X login/browser automation blockers (it switched to chrome dont know why, i use Brave browser )
- wrong draft focus
- body paste accidentally targeting title
- too many blank lines
- Markdown - item not becoming real X bullets( solved after at lease mentioned it twice )
- Preview button unreliable
Human helped by logging in, clicking the right draft, and spotting formatting issues with screenshots.
Codex fixed it by:
- collapsing extra newlines
- converting Markdown lists into native X lists
- removing leftover -
- using direct /preview
- updating the reusable skill

[Wade4AIWeb3 @no7wade]

@OpenAIDevs usage

[Wade4AIWeb3 @no7wade]

@OpenAIDevs orginal markdown looks like , the article result https://t.co/eoKvKLTaQH , work now ,though during the process still need to debug and improve again and again.

[Wade4AIWeb3 @no7wade]

https://t.co/v9FJSQmbWX

[Wade4AIWeb3 @no7wade]

which means that you still need to use LLM to help (cost u token of cause, and wont finish instance as script, apart from those waiting gap) , in another world, more capable to handle exception.

[Wade4AIWeb3 @no7wade]

AI + RPA / Computer / Browser use is eating XYZ’s market share.
Only a few moats matter for agent startups:
- Proprietary data
- Regulatory moat
- Deep workflow integration
- Integrated hardware-AI

[OpenAI Developers @OpenAIDevs]

Show Codex a workflow once. Reuse it as a skill.

Record & Replay lets you show Codex a recurring task, like filing an expense report or submitting a time-off request.

Codex turns that demo into an inspectable, editable skill.

You control when recording starts and stops.

[Wade4AIWeb3 @no7wade]

Another critical distinction: skill ONLY generated not scripted that means true semantic adaptability, zero/lower maintenance debt, and great composability.

[Wade4AIWeb3 @no7wade]

https://t.co/v9FJSQmbWX

[Wade4AIWeb3 @no7wade]

https://t.co/OZ5MN3rN4G

[Wade4AIWeb3 @no7wade]

AI Coding Time Report
- npx cchour
or for english and better category
- npx github:no7dw/cchour --lang en
credit @jianshuo 🙌