runesleo

vip
Active for: 8.6y
Peak Tier 6
No content yet
Connect this 8-core Grok Bot machine to your own AI. Use SSH.
8 cores, 15G, Debian. The regular port 22 is inaccessible. I didn't set it up manually; I had AI establish access with Tailscale SSH. It keeps running after I close my laptop. Kill the network, and it’s back online in about 20 seconds.
The page is just the entry point for provisioning the machine. Once SSH is working, your usual Cursor and Claude can use this high-performance machine like a real computer.
For your use case, just send this to your AI:
Connect this Grok Bot cloud machine to my computer so my own AI can operate it via
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
Whether you’re happy or not has little to do with what you’re doing.
Whether you’re mentally present while doing it matters a lot.
This was a 2010 Harvard study published in Science, using random iPhone pop-ups to sample the real-time states of more than 2,000 people in daily life.
—— Three numbers ——
 People’s minds wandered during 46.9% of their waking hours
 “What they were doing” explained only 4.6% of the variation in happiness
 “Whether their minds wandered” explained 10.8%
In other words, what you’re thinking about affects how happy you feel in the moment more than twice as much as what
View Original
  • Reward
  • Comment
  • Repost
  • Share
You can have written cleanup code, and that cleanup code never having executed, both be true.
And no errors are reported at any point.
—— How it was discovered ——
A long-running background process, with all the temporary files that should have been cleaned up still there every time. They kept piling up.
Looking through the code, the cleanup logic was written properly and registered with atexit.
It looked completely correct.
—— The truth ——
atexit only runs when the process “exits normally.”
And long-running processes almost never exit normally. They usually end up:
 Being killed (SIGTERM by de
View Original
  • Reward
  • Comment
  • Repost
  • Share
I copied a file as a backup.
Both copies ended up being useless.
The reason: that file contained not data, but an identity.
Copy data once, and you have two sets of data.
Copy an identity once, and you don’t have two identities — you have the same identity appearing in two places at the same time.
Any platform that sees this will first assume, “The account has been hacked.” So the entire credential is invalidated, and the copy and the original die together.
There are more things like this than most people think:
· Account login sessions
· API keys, authorization files, certificates
· Browser c
View Original
  • Reward
  • Comment
  • Repost
  • Share
I just received $599.
It’s not an ad deal, and it’s not freelance work. Someone read what I wrote and came over on their own to buy it.
Here’s something I’ve never talked about publicly—
My paid content has sold more than thirty times to date. The $49 long-form paid articles, the subscription section on X, plus this order.
Not a single one came from a landing page.
Not one. All + USDC.
For a while, I thought the right way to sell paid content was: build a good landing page and optimize the conversion rate.
The data told me otherwise.
What really leads to a purchase is that the person has alrea
USDC0.00%
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
Today I went back to my hometown, and for the first time in all these years, I didn’t bring a computer with me.
That used to be impossible. Whether traveling, visiting family, or even just going out, I either had to carry a computer so I could handle things at any time, or simply assume that nothing would get done that day.
Now one phone is enough.
Before leaving, I told the AI on my phone, “Confirm that the computer is in working condition.” It checked everything that needed checking and sent me a list. Then I left.
The things that are actually running were already on the server, so they have
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
I always thought Codex had been absurdly slow lately because I had installed too many Skills.
Then I checked the logs, and that was completely wrong.
One step took 138.6 minutes, involved 206 model/tool round trips, and spent 54.9 minutes just waiting for sub-Agents—but it didn’t load a single Skill.
What really slows down Agents is often neither the model nor Skills, but over-orchestration:
- Stacking full-history sub-Agents under the Ultra controller
- Multiple rounds of “review → fix → re-review”
- High-frequency polling
- Running full-repository tests for every minor change
I’ve just chang
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
When you work alone, the hardest part isn’t having nothing to do; it’s opening your computer in the morning and not knowing where to start.
I later realized the root of the problem was trying to do two completely different things at the same time.
One is **looking at the world**—what’s happening out there, whether anything relevant to me has changed, and which developments are worth following. The nature of this task is filtering: the input comes from outside, and when it’s done, there should be a list of “0–3 things worth looking at.”
The other is **managing yourself**—what I have in progress
View Original
  • Reward
  • Comment
  • Repost
  • Share
When you ask AI to handle something for you, how does it get past the password step?
I’ve seen three approaches, and the risks are worlds apart:
1. Paste the password directly into the conversation—the easiest, and the worst. That sentence will enter the context, the logs, and possibly the training data, and you can’t take it back.
2. Have the Agent read your password manager—it sounds professional, but it gets the plaintext, and you don’t know where that context goes afterward.
3. The Agent only initiates the request; you click to confirm in the password manager’s own interface, and the value
View Original
  • Reward
  • Comment
  • Repost
  • Share
Super convenient automatic replacement of pinned conversations in an AI agent client! The conversation chip in AI apps is so useful now.
When a conversation goes on too long and reaches its limit, there’s no need to force it or switch manually.
Just have the AI automatically open a new window with the same role and continue the conversation,
or replace it with a new sidebar entry pinned at the top.
One automation, three uses:
1. Direct it in this conversation to take over another conversation as the pinned one
2. Open a new window to continue with forked or temporary tasks
3. Dis
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
Secured the Cloudflare Wallet brand name first:
☁️
Currently, each Cloudflare account can claim only one handle. If you want to keep the brand name, you can go take a look first:
View Original
  • Reward
  • Comment
  • Repost
  • Share
Try this prompt with your own AI:
【Discuss with me what kind of person I want to become ten years from now.
Don't just talk about abstract visions; help me clarify my finances, work, health, and relationships.
From now on, guide me through a weekly review: Did what I did this week bring me closer to that direction? If I veered off course, tell me the one thing I should adjust next week.】
View Original
  • Reward
  • Comment
  • Repost
  • Share
Many people misunderstood this change:
Polymarket didn’t change the starting line; it changed the end-point determination.
The starting line is still this round’s Price to Beat; the end point has changed from the last instantaneous price to the average price over the last 30/60 seconds.
View Original
  • Reward
  • Comment
  • Repost
  • Share
A Polymarket player with about a 96% win rate on the main battleground: I split BipBop’s 1,632 trades into separate positions
This account on the leaderboard: BipBop
1,632 settled positions, main battleground win rate about 96%, positive for multiple consecutive months.
Most people’s first reaction is: copy him.
We actually backtested that layer that looks “like him” — turning “buy 0.85–0.95 high-probability markets, and get to settlement” into a mechanical strategy, backtesting 5,378 markets (high-probability bucket compared group): win rate 88.9%, net profit −2.5%.
Win rate is an illusion. A
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
Codex is down—congratulations, everyone is about to welcome a new round of credit limit resets!
View Original
post-image
  • Reward
  • Comment
  • Repost
  • Share
Automated trading is all the rage right now, and everyone is teaching people to have Claude place orders directly.
But permissions shouldn’t be tied together like this.
One rule you can take away is: research permissions can be given to an Agent, but trading permissions (wallet, signing, placing orders) must not be handed over together.
The correct split is: only open a public, read-only research interface; the execution interface is always separate and requires manual approval.
polymarket-toolkit does exactly this.
After GitHub open-sources runesleo/polymarket-toolkit, MCP provides
View Original
  • Reward
  • Comment
  • Repost
  • Share
When you’re not at home and the elderly are not by your side, and they also have very specific needs, with AI now, it really becomes convenient to create highly customized small tools.
For example, in my family, an elderly person needs to do muscle rehabilitation training for a period of time. I turned the movements and intensities that have already been confirmed into a web page specifically for him: when you open it, you can immediately see what to train today, with one item per screen that you can follow directly. If the plan changes later, I can also update the page for them right away.
In
View Original
post-image
post-image
  • Reward
  • Comment
  • Repost
  • Share
  • Pinned