I sometimes ship open source. Harness engineering for aliens.
Singapore
Joined January 2022
- Tweets3.3K
- Following1.3K
- Followers988
- Likes6.4K
Claude Code started reading AGENTS.md when there’s no CLAUDE.md.
I used to keep two instruction files in sync for Codex and Claude. That’s a tax. I’m deleting it.
Shared repo rules in one place. Vendor files only for the leftovers.
New card (cream / split / less funeral):
**Copy (warmer):**
I keep splitting CLAUDE.md the wrong way.
Taste stays in prose — house style, patterns, “we usually do X.”
Anything I can’t afford to miss leaves the prompt: secrets, push to main, done without tests. Those become PreToolUse / Stop gates.
Preference in words. Invariants in gates.
The best tools are the ones you keep running out of.
Grok's agents are brilliant. They're also the fastest way I've found to hit a usage cap.
Intelligence got cheap. Meanwhile quota..
I ran the same lockthree brief on Claude Code and Grok Build.
Both hit correctness 100.
Claude: 233s · review 1
Grok: 461s · review 2
Same task. Same brief. Claude was nearly twice as fast. Both shipped a working CLI.
Rule I keep: score what the brief requires — not a harness flag the brief never asked for.
Polymarket has successfully evolved from a niche application into a formidable collective intelligence engine, attracting a high-value, rational user base that Web3 has historically failed to capture.
Seeking connections with Polymarket enthusiasts and builders.
DMs open for those shaping the future of prediction markets.
The key to Polymarket isn't chasing smart money or insiders but spotting consensus among those wallets.
passive earn at Bitcoin Price on Nov 1 market.
we've already seen the volatility in BTC markets on Polymarket.
but look at this from another angle.
here's my offer:
buy the outcomes in exact range and hold till profit comes to you.
look at tendency:
Oct 27 price: ~$115,458
Oct 20 price: ~$110,000 (approx from trends)
Oct 13 price: ~$105,000
Oct 6 price: ~$125,000
Sep 29 price: ~$120,000
you can see that BTC fluctuates around $110k-$120k lately, with average weekly change of 5-10%. but predicting only $112k-$118k may make you lose.
then you expand outcome range buys and calculate profit:
Oct 28-Nov 1 market. (wait, Nov 1 price market)
launched recently, current price is ~$115k.
first few days had some volatility due to weekend news.
BTC hasn't moved as much as during major events.
the odds already reacted and we see $114k-$116k at the top ~44¢.
so, I can predict the price staying stable or slight increase and make buys of these outcomes:
(110,000-112,000) + (112,000-114,000) + (114,000-116,000) + (116,000-118,000)
total cost of odds calculated as 5+28+44+18 = 95¢.
if any of outcomes I bought will be resoluted, I get ~5% profit to my total position.
okay, to feel myself more confident I add (108,000-110,000) + (118,000-120,000) outcomes.
and even after this I get a guaranteed >0% but safe coverage.
4 days. decent APY if scaled.
Nov 1-Nov 8 market.
trades already started. thinking from tendency of recent prices. odds already help us:
(106,000-108,000) + (108,000-110,000) + (110,000-112,000) + (112,000-114,000) + (114,000-116,000)
total cost of odds are calculated as 70¢ (assuming adjusted).
43% profit in 8 days. or ~1,900% APY.
the main thing.
these calculations are set for the worst type of traders - they aped and forgot.
while you can always check the market tendency, be alerted to BTC price changes and remove the piece of range that might be out soon.
the closer end of the market, the higher odds total cost. so we try to enter earlier.
links at Bitcoin Price markets in replies.
Will inflation reach more than 3% in 2025?
NO is so free here
polymarket.com/event/how-hig…
