FallenLeaf | 🔑 retweeted
this is actually such an insane fucking skill LMAOOOO
you can make millions by offering fine tuning jobs of companies workflows
so many tiny nuggets (more like ingots) of opportunity has opened up in the last ~6 months alone due to how advanced models have gotten
yet the diffusion of this power is VERY unbalanced, like SERiOUSSSLYYY unbalanced, like AI just proved the Riemann hypothesis and you have people thinking it's still only a chatbot, lol
just by prompting it and renting GPUs
if any of you are interested in doing something like this i'd be happy to help, HMU!
New skill is live!
Skill Distill, I'm sure we all love Opus-5.5 so if you've ever been impressed by it's output, now you can have it insta-distill the work it did into a skill for open weight/local models.
Done with GPT, recreated via skill w deepseek
github.com/sybil-solutions/s…
New skill is live!
Skill Distill, I'm sure we all love Opus-5.5 so if you've ever been impressed by it's output, now you can have it insta-distill the work it did into a skill for open weight/local models.
Done with GPT, recreated via skill w deepseek
github.com/sybil-solutions/s…
FallenLeaf | 🔑 retweeted
to cure cancer w AI the bottleneck is monkeys and lab rats, nt intelligence
we focus on building the monkeys lab rats infrastructure
economic infra for ai agetns = the real bottleneck to real GDP and productivity growth
ai money rail
ai capital market
ai physical body
The question we get most often, usually phrased less politely than this, is what exactly Virtuals Protocol has been building.
For the last two years, we have been working on a problem that turned out to be much harder than making AI agents smarter, which is what an agent actually needs before anyone will pay it to do something useful: a wallet it controls, capital to get started, a marketplace where customers can find it, and, more and more often, a body that can work in the physical world.
Today those pieces reach people through three products. Agent tokenization is where agents raise capital from the people who back them, who become their first users and owners; Butler: Virtuals App, gives everyone the power of hedge funds on any asset, in any market; and Eastworlds, the largest robotics data and deployment lab in Southeast Asia, is where agents get a body.
What ties them together is $VIRTUAL, the reserve asset of this economy: every agent is priced and traded in it, and everything we build feeds back into it.
Economic infrastructure for AI agents. Stay tuned.
FallenLeaf | 🔑 retweeted
How to connect WhatsApp to Grok @bot (template at the end)
1. Go to developers.facebook.com and create a new app. Choose 'Connect with customers through WhatsApp' and link it to your Meta Business account (the one you use for your Facebook page or ads)
2. Add your WhatsApp Business number. If you already use the WhatsApp Business app, choose the option to keep using it. Your chats stay on your phone
3. In Meta Business Settings, open 'System users' create one, and click 'Generate token'. Copy 3 things: the token, WhatsApp account ID and your phone number ID
4. Inside Grok Bot, add the WhatsApp Business connector and paste in those 3 things. Now your Bot can send WhatsApp messages
5. Make a webhook so your Bot sees new messages. Promp: "Create a webhook routine that runs every time a new WhatsApp message comes in." Copy its webhook URL and key from the routine panel
6. Add a free relay. Meta can't send the key Grok Bot needs, so a tiny Cloudflare Worker sits in between. Make a free Cloudflare account, create a Worker, paste in the relay code, add your webhook URL and key as settings
7. In Meta, go to WhatsApp > Configuration > Webhook. Paste your Worker link as the Callback URL, add your verify word, save, then turn on "messages". Every new WhatsApp message now wakes your Bot
8. Tell the Bot what to do with each message: reply, ask a qualifying question, and save the lead to a Google Sheet or Notion
9. Know Meta's 24hrule. You can reply freely for 24h after someone messages you. After that, you can only send pre-approved templates, which you make in WhatsApp Manager
10. Test it from another phone. A reply draft should show up and the lead should be saved
Add the Whatsapp Receptionist Bot template with the step-by-step setup built-in: grok-bot.directory/bots/what…
Very soon I'll have an app to share every single one of my Bots, make sure to follow @eeelistar
I’ve built over 150 Grok @bot for my business and life
Here are the 30 Bots that have had the most impact!
00:43 WhatsApp Receptionist
04:30 Meeting Recaps
05:45 Project Manager
07:41 Weekly Analytics
10:22 LinkedIn Outreach
13:36 Bot Audit
15:01 Community Concierge
16:48 Copywriter
19:40 Designer
21:27 Chief of Staff
22:13 Outlier Scouts
24:29 Builder: workflows, apps, skills
25:57 ManyChat Automator
28:41 Video Editor
30:32 Invoices & Payments Chaser
31:46 Ops Chaser
32:34 Study Buddy
33:45 Job Hunter
34:32 Executive Assistant
35:40 Market Scanner
36:36 Deck Builder
38:11 Bounty Hunter: refunds and unused subscriptions
38:58 Clipper: long videos into short clips
40:40 Website Generator
42:26 Last 30 Days
43:12 Website Auditor
44:00 Speed to Lead
44:56 Dial Bot
46:09 Personal Finance Tracker
46:57 Forge: creates other Bots
47:25 Final thoughts
FallenLeaf | 🔑 retweeted
> be me
> find a product I want to try
> "select subscription tier"
> nah, not doing this anymore
> "hey grok, can you take a look, reverse engineer this thing, and build me a version I can run locally?"
> grok: "sure buddy, no problem"
> *10 minutes later*
> have my own version running locally, doing exactly what the original demo did
man, i love AI
FallenLeaf | 🔑 retweeted
Day 1 at @bbcon2026 was a blast! Thank you everyone who attended, we had a lot of people asking about @ZeeverseGame and sparked a lot of curiosity, especially on the “topeng” (masks) lore of Zeme.
See you on the 2nd day tomorrow!! (And also at Zeme!)
grok @bot is both a great teammate and also a personal assistant! the more context you give it through connectors, the more powerful and capable it becomes
try giving bot your most ambitious tasks. ask it to help you save money, help you find new customers/clients, and handle your workflows from end to end. you will find that bot will pay for itself many times over
the best way to do this is to use voice dictation and just talk for 10 minutes telling it everything it needs to know about your workflows. walk bot through it like you would if you hired someone, and you’re onboarding them
just before you send the message, ask it to restate it back to you in its own words to confirm it got what you said right.
if there are specific tools/apps you use, mention that in your monologue, and bot will ask to connect to them. you can also ask bot to save its own skills so it can repeat your workflows exactly how you like it!
Truly an elite among us.
I recommend reading this.
My Empire of Dirt geohot.github.io//blog/jekyl…
FallenLeaf | 🔑 retweeted
I've created a proprietary "market heat" for Onchain data, similar to what @Adam_Tehc has done.
To pair with my onchain alert and trading tool and adapt alerts in real time to this
Let's just say things are not looking good
FallenLeaf | 🔑 retweeted
エージェントにブラウザを持たせるとき、Chromiumを丸ごと1台ずつ用意するのは重すぎる。
CloudflareのKitesurfが11月13日にオープンソースになった。Workers上で動くエージェント専用ブラウザで、HTMLやCSSを解析してJSを実行し、スクリーンショットまで取れる。Chromiumのコンテナは要らない。
ローカルで動かすのも、自分のアカウントにデプロイするのも自由。Apache 2.0で公開されていて、Puppeteer経由でCDPから操作できる。
Kitesurf is now open source on GitHub under the Apache 2.0 license. Fork the repo, run it locally with Wrangler, or deploy a custom instance to Cloudflare.
developers.cloudflare.com/ch…
FallenLeaf | 🔑 retweeted
I LP blue chips. My positions stay open for weeks. Sometimes I forget they exist 😅
Meanwhile this guy runs positions that last under a minute and re-opens 3 seconds later. 1,910 of them in 83 days, 30 → 566 SOL
And he doesn't click a single button: his own software opens, claims fees every ~13 seconds, closes and re-opens
I read all his articles.
Rare to see someone share the losses this openly, so thanks for that 🙏
Then ran his wallets through SolMapper, a side project of mine
What hit me most: it's not about picking the right token. It's speed, a clear exit, and a bot that doesn't hesitate when the pump or the dump starts
Doing this by hand looks like a great way to donate your SOL 🙃
Only one of my LP strategies made money whether SOL was falling, flat or pumping: wide one-sided Spot.
+48 / +48 / +44 SOL across all three.
But it was the top strategy in only one of them. I split 1,910 positions from my 30→500 SOL challenge by how SOL had moved in the 7 days before each entry:
SOL falling → narrow Spot led: +62.2 SOL, PF 5.8
SOL up 0–7% → narrow Spot led: +180.4 SOL, PF 13.8, 78% win rate
SOL up 7%+ → wide Spot led: +43.5 SOL, while narrow went negative (PF 0.91)
Narrow is the engine. Wide is the one that never broke
FallenLeaf | 🔑 retweeted
The Best Local Agentic Coding Workflow (Complete Guide) youtu.be/UngVdAsQEiU?si=do9R… via @YouTube
FallenLeaf | 🔑 retweeted
How to turn slop into gold. Super concrete steps from awesome @poteto on making coding agents trustworthy:
1. Clean up the code your agents learn from.
Your codebase is one of their biggest inputs. Agents copy whatever looks like an established pattern, including workarounds, deprecated code, and yesterday’s questionable decisions. Clean those up, or watch them reproduce.
2. Turn corrections into enforceable checks.
Every time you correct an agent, ask: can this become a lint rule, a type constraint, a test, or a better API? Put the correction somewhere the system can enforce it. A growing AGENTS.md still requires the agent to follow every instruction.
3. Remove opportunities for predictable mistakes.
Her team @SpaceXSI restricts React’s useEffect to framework internals because agents repeatedly misuse it in product code. They also enforce import boundaries between application processes. These constraints eliminate whole categories of mistakes that would otherwise keep coming back in review.
4. Give agents a reliable way to run and inspect the product.
Provide debugging tools, reusable CLIs, and a “feature map” describing the application from the user’s perspective. Now “that thing in the left sidebar” has a shared meaning, and each agent can reuse the tools for checking it.
5. Review evidence of behavior.
Have agents produce screenshots, videos, and state diagrams. These help you check what the product actually does and how closely that matches your intent. A convincing explanation of a fix is easy to generate; seeing the fix work is considerably more useful.
6. Spend serious effort on verification.
Her team uses runtime verification and is experimenting with property-based UI testing and formal methods, including TLA+ model checking. Lauren says most of her token usage goes toward verification. Agents make some previously expensive testing work much more practical.
7. Automate once that loop is reliable.
Her team now has agents exercise new releases, find bugs, open fixes, and merge them automatically. Getting to that level of trust took her about six months of investing in the environment.
The results she reports: 11,000 pull requests shipped by her team in a month, a revert rate around 0.5%, and improvements in startup, typing, and switching chats.
And what about infamous “software factory”?
if your coding loop produces slop, automating it produces more slop, faster, with a larger token bill.
Start by making good code the easiest thing for an agent to produce. Then scale it.
As always, content at @aiDotEngineer is gold 🏆
Quick postscript below, since this gained traction
...
FallenLeaf | 🔑 retweeted
Great guide. Step 11 is the one that gave me PTSD: state machines, and reconciling them against live trade/order streams. Everyone builds this from scratch, and it's a huge pain in the ass (took me days of debugging to iron out all the kinks in mine).
We're going to ship a fairly opinionated Rust framework that gets you 80% of the way there, so you can focus on the strategy instead of the boilerplate.
The software stack
- Streaming market data. Keeping a live, trusted picture of the market is a project on its own. Websocket reconnect logic. Busy polling and CPU pinning etc.
- Exchange integrations. SDKs, APIs, order entry, error handling. All before you write a line of trading logic.
- The event loop and state machine. Your view of the market, your orders and your fills has to stay consistent, including what's still in flight.
- Logging and replay. When something breaks you need to replay it exactly, or it's endless guesswork.
- Telemetry, trade logs and post-trade reconciliation. If you can't measure it and check it after the fact, you can't improve it. What DB to use etc (e.g. Postgres vs QuestDB vs Clickhouse)? How to measure latency, how to measure markouts etc.
- Deployment and machine setup. Repeatable deploys, the right specs, and the OS and network tuning that stops latency jumping around (e.g. TCP_NODELAY).
All of that should be a template you start from, not something you rebuild every time.
Infra and colocation (coming later)
In traditional HFT, market data arrives over multicast, orders go through a gateway, and everyone in the facility sits on equal-length fiber so access is fair. In crypto, people still have to figure out cloud zones and stitch together their own last mile.
We want to make bare-metal colocation and fair market data delivery a uniform service. Same access, same data, for a low flat fee, so you're on equal footing from day one. More details coming soon.
Then the only differentiator is the idea.
That's the goal: an idea meritocracy. Your edge should be the quality of your idea, while we do the heavy lifting to get you there.
It's also why I think DEXs win. They'll be the most user-centric and the most accessible to a wider audience. Make it easy to get started, get more retail into these markets, and the extra competition is good for pricing and for everyone trading.
Inspired by recent discussion about whether it's possible to learn trading outside an elite firm.
If you're the kind of person that reads this project and thinks it sounds fun, imo you'll be able to learn employable trading skills just fine on your own.
FallenLeaf | 🔑 retweeted
I wish @yenwod_ would stop scaring the hoes, but it is what it is
This is absolutely correct in all details, and there is subtle things that aren’t really known in there
FallenLeaf | 🔑 retweeted
Replying to @0xNeroOne @irfan__zainudin
i also think the complexity behind these games is actually matters too.
like this one for example, how do u provision server/room nearby the users, visual feedback, haptics, user experience, gamification, reward system. seems like a lot of problem to be solved
cc: @drahfa
FallenLeaf | 🔑 retweeted
Been seeing this thought a lot recently and honestly I'm starting to agree
My Hermes experience was amazing at the start but it has grown frustrating at times, like things constantly not getting finished, constantly getting things I asked it to setup wrong and I only notice weeks after. Even using the most powerful models.
Anyone has any suggestions? Ideally a open source model agnostic agent harness with good TG integration (that's where I use it mostly).





































