Закреплённый твит
m0h
56.7K posts

m0h
@exploraX_
ai research & web3
learning agentic automation... Присоединился Eylül 2019
654 Подписки19.2K Подписчики


here’s a video guide on how to use claude to create your own agentic skills.
in this video, i asked claude to build a meta-skill that creates other agentic skills just by asking simple questions.
then i put the skill to the test, and the result was incredible. i’ll leave the prompt i used in the comments.
m0h@exploraX_
English

Arc just dropped two major updates.
And they’re easy to miss if you’re not looking closely.
First, ERC-8183 is now live on @Arc testnet. I know it sounds boring, but trust me, it’s powerful.
Second, they just launched their ambassador program, and here’s how to get selected:
> They’re selecting people manually
> Based on your contribution to the Arc ecosystem
- Create content.
- Break things down.
- Build.
- Stay consistent.
Now back to the first update.
ERC-8183 introduces something called “Jobs.”
In simple terms:
AI agents can now do tasks for each other and get paid onchain.
The payment is locked first. Then released after the work is done.
So no one gets cheated.
That’s how agents can actually work and earn onchain.
But working together is not enough. Money also needs to move easily.
That’s where Arc comes in:
- You can use USDC as gas
- You can pay with USDC or EURC
- It handles currency conversion for you
So in simple terms:
ERC-8183 is how agents work while Arc is how money moves
Arc@arc
Introducing Arc House, the new home for the Arc community, and Architects, the program recognizing the builders shaping the Arc ecosystem. Arc House brings everything together: → Educational content and discussions → Hackathons and events → Community groups and meetups → Builder activity and recognition The Architects program introduces a tiered system where contributors earn points, unlock roles, and gain visibility across the ecosystem. Arc House is where the community gathers. Architects recognize the people building it. arc.network/blog/introduci…
English

10 Claude Code tricks most people don't know (save this):
1. CLAUDE.md: permanent memory. Claude reads it every session. Stack, rules, naming conventions. Never explain yourself again.
2. /clear between tasks: context degrades at 90%+. Claude gets measurably dumber. New task = /clear. Every time.
3. Esc Esc: checkpoint menu. Rewind code + conversation to any point. Zero damage if Claude derails.
4. claude -w feature-branch: isolated git worktree. Claude works there, commits, PRs. Main never touched.
5. git diff main | claude -p "review": pipe anything into Claude. Logs, diffs, files. -p = non-interactive mode.
6. --effort high: 4 levels: low / medium / high / max. Default is not max. Use high on hard problems.
7. Auto memory: Claude saves its own learnings across sessions. Build commands, quirks, your preferences. Zero effort.
8. --max-budget-usd 5.00: hard spend cap per session. Pair with --max-turns 3. Essential for CI/CD.
9. .claude/rules/api.md: path-scoped rules. API rules only load for API files. Test rules for test files only.
10. claude -n "auth-refactor": name your sessions. Resume with claude -r "auth-refactor". Clean parallel workstreams.
Most people use 2 of these 10.

English

@bloomstarbms @Cryptocolltex that’s a valid question, because there’s a difference between I’m into a skill and learning a skill
English

@damianplayer @therituallab this feature is called Hook in Claude docs, basically you’d set a background control for Claude instances.
once an agent is done with a task, it would play a sound, or display a Colour the way the guy is or animations to tell the user that the agent need the user attention
English

@exploraX_ Thanks, exactly what I needed for my daily routine
English

how to become an automation engineer in a day (no coding experience):
> last year, running ai agents autonomously required some level of coding.
> perplexity AI changes that, making autonomous agents accessible to anyone.
> go to perplexity.ai and set it up by connecting your tools first.
> click “connectors” in the sidebar. it supports 400+ apps like gmail, google drive, slack, etc.
> brief it on who you are. for example:
“i’m a [role] in [industry]. i regularly produce [x, y, z]. remember this for every session.”
> show it what good looks like, find 2–3 examples of your best work.
> upload them and say:
“these are examples of my best work. study the format and tone. use them as a reference every time you create output for me.”
> now start assigning tasks.
> the better your prompt, the better the output. treat it like you’re delegating work to an employee.
> example (marketing):
“analyze top-performing content from [competitor 1], [competitor 2], [competitor 3] over the last 30 days. identify formats and topics driving engagement. find gaps. build a 30-day content calendar based on those gaps. save as a google doc.”
> set it on a schedule: get a fresh competitive brief every monday. no manual research.
and that’s it.
cc: @damianplayer
Damian Player@damianplayer
English

@exploraX_ That 'complete terms in license.txt', is it suppose to be part of description ?
English

here’s a video guide showing how to easily add any agentic skill to Claude.
the article includes a list of 20 powerful agentic skills you can use with claude.
save this post for later.
m0h@exploraX_
English

@exploraX_ Brother, I understand that learning AI stuff is very useful for personal use
But are there companies or organisations that hire people for having that skill?
English

@Albright_MXM yes you can sir, I’ll post a guide on how to create your Claude skill tomorrow
English

@exploraX_ Can I use Claude to build a trading bot that can trade tokens, stocks and commodities 24/7 based on market sentiments
English

🚨BREAKING : SEC just announced that any token without at least 20% airdrop to community users will be classified as a security.
🔗 Source : google.com/url?q=https://…

YouTube

English







