One who can make u day
Joined October 2024
- Tweets4.6K
- Following357
- Followers2.6K
- Likes5.6K
Dark retweeted
THIS IS A REAL F*CKING GOLDMINE
Anthropic engineers just revealed the most powerful multi-model pipeline for Claude Code
The move: chain Sonnet, Haiku, and Opus into an automated conveyor where each model handles only what it does best
Maximum development velocity with zero context waste: lightweight models do the heavy lifting, while the frontier model drives architectural strategy
You just launch Claude Code like this:
CLAUDE_CODE_SUBAGENT_MODEL=haiku claude --model sonnet
here's how it splits:
→ Sonnet 5.5 drives the primary session, writes clean code, and directly applies diffs
→ Haiku 5.5 swarms the codebase in parallel: instant grep scans, call-site tracing, and spec retrieval
→ the Opus 5.5 architect subagent oversees overall system design and signs off on the final diff
→ you stop drowning your expensive frontier model in thousands of lines of raw terminal logs
Opus only steps in at three critical moments:
→ before locking a multi-file architectural plan
→ when a test breaks twice at the exact same failure site
→ before staging the final commit and assembling a pull request
Sonnet builds · Haiku scans · Opus reviews
Prompt cache preservation under the hood: Sonnet's primary session keeps its prompt cache prefix active at a 90% discount, while Haiku subagents spawn in isolated working environments and dissolve without polluting your main conversation window
the full tree
> Sonnet 5.5 runs the primary session and writes surgical code edits
> subagent-search maps repository structure on Haiku 5.5 in read-only mode
> subagent-grep pinpoints function call sites and returns concise structured summaries
> subagent-audit verifies test diffs inside a temporary worktree without touching your branch
> subagent-architect on Opus 5.5 approves the final pull request
> hard automated ceilings abort looping diff attempts after two consecutive failures
paste the tree and this prompt into Claude Code ↓
"Reconfigure my Claude Code setup around this tree:
• Run the main session on Sonnet 5.5 and set CLAUDE_CODE_SUBAGENT_MODEL=haiku in ~/.claude/settings.json
• Create specialized subagents in .claude/agents/: search, grep, and audit on Haiku 5.5 with tools: Read, Grep, Glob, plus architect on Opus
• Configure worktree isolation for test runners so temporary edits never pollute active branches
• Add one rule to CLAUDE.md: delegate all codebase exploration to Haiku, summon architect after two identical test failures, and enforce diff audit before git staging
Show every change as a diff first. No edits until I say go"
code.claude.com/docs/en/sub-…
save it, you'll need it
Dark retweeted
Larry Ellison has now cancelled his plan to sell up to $7.5B worth of $ORCL stock.
No shares were sold, and Oracle says he has no other plans to sell any of his stock.
$ORCL is currently trading around $150, down more than 50% from its record high of $345.
Oracle founder Larry Ellison is planning to sell up to $7.5B of $ORCL.
His new trading plan allows him to sell up to 50M shares through October 24.
Even if he sells it all, he would still hold around 1.1B shares, more than 40% of Oracle.
Dark retweeted
A GROK ENGINEER JUST SHOWED HOW TO MAKE AGENTS FORGET ON PURPOSE AND CUT CONTEXT COST BY 90%
most agents die from remembering. 200 tool calls, 40 file reads, 3 failed attempts - all of it still sitting in the window, still being paid for on every single turn
the fix runs in 2 passes → one agent does the work in a scratch context → a second pass keeps only the 4 things that survived: the decision, the file paths touched, the constraint that broke, the version that shipped
everything else gets dropped. not summarized. dropped
the raw run lives on disk, not in the prompt. the agent reads it back only when a task actually points at it, so a 180K-token session compresses to a 6K-token handoff with nothing load-bearing lost
then the next run starts clean and still knows what the last 12 runs cost
most people scale agents by feeding them more. this one scales by deciding what to throw away.
Dark retweeted
One dude on Polymarket turned $100 into $4,923 pure profit…
just by betting on the weather
2,000+ trades, $19k+ total winnings — almost all from daily temperature markets in random cities.
His stupidly simple edge:
• Snipes YES at 5–15¢ (when the market sleeps on obvious heat/cold)
• Grabs NO at 40–55¢ (when the crowd overreacts)
• Never risks more than $1 per bet
• Spams 30–50 micro-positions every single day
Result? Near-perfect win rate. Biggest bags:
• $119 → $4,923 (+4,033%!)
• $109 → $949
• $33 → $536
• $4 → $499 (+12,035% holy shit)
Is weather on Polymarket still this broken forever?
Or are bots + whales about to close the gap and kill the free money?
Check him: polymarket.com?via=neobrothe…
Dark retweeted
$120k in one day for the bot — this is just the beginning, thanks to
( polyguess.fun/?ref=H15vRhwoQ… )
In just one day, I managed to raise this amount through subscriptions.
The winter promotion for the bot is now live, plus a small bonus from the team.
Dark retweeted
This insane AI bot on Polymarket turned just $300 into over $115k in only a few days!
It’s crushing it almost exclusively on the super-short XRP 15-min Up/Down markets.
The play is brutally effective:
It scans fresh markets in real time → spots tiny inefficiencies or mispricings → loads up on the high-probability side (or both when the edge appears) → rides to resolution for locked-in gains.
Insane stats: 123 trades, 87 winners → ~70.7% hit rate. Average bet ~$4k. Pure machine precision.
Check the monster yourself: polymarket.com?via=melfoy
Still grinding manual trades in 2026?
What’s really holding you back from coding your own beast?
Dark retweeted
Suspicious bets around Iran are popping up on Polymarket.
New accounts are placing bets with very low odds - under 18%.
They’re betting on US strikes before January 31, 2026 and on Khamenei stepping down.
I think I just found a potential poor Iranian insider
Profile: polymarket.com/@0x4128Be7113…
Brand new Polymarket account ~4 hours old and it instantly dumps $6,360 into YES on Khamenei out by Jan 31
That’s way too small for a real insider…either it’s a test transaction?
Dark retweeted
Bot exploded $2K into $357K on @Polymarket in just one month
Custom bot generates ~$150K monthly revenue, pure mechanical edge
Scalps 90¢-99¢ contracts + auto-compounds 27K+ micro-positions daily
Captures spreads + LP incentives with zero emotional interference:
Expected directional gambling, discovered systematic LP scalping instead
Master trader exploits Polymarket's core mechanics for high-probability scalps + liquidity rewards
> Buys 90-99¢ contracts on near-certain resolutions repeatedly
> Short-term crypto Up/Down markets scalped for consistent micro-gains
> Auto-compounds thousands of positions while harvesting spreads
Profile: polymarket.com/@CRYINGLITTLE…
