54 Photos and videos
I built @PolicyPoolHQ to answer one question:
what if liquidity could say “not at that size, not unless LPs get paid first”?
v4 Hooks make that enforceable.
Same order. Two pool policies.
One reverts. One pays LPs, then clears.
This is why hooks matter.
Jun 4
Uniswap V4 Hooks Winners:
🥇 First Place - @Eulr_Official
Eulr is a fair token launchpad with no presales & team allocations, where onchain rules prevent price crashes.
🥈 Second Place - @HatchAI_App, @PolicyPoolHQ
HatchAI is token launchpad that automatically protects new projects from sniper bots and whale dumps through built-in pool-level safeguards.
PolicyPool is a trading pool with programmable rules, where oversized trades either get blocked or are routed through a mechanism that rewards liquidity providers first.
🥉 Third Place - @FanVibeOnX, @4lpha_agent, @RouletteSwap
FanVibe Hook is a World Cup prediction market that connects fan activity directly to DeFi liquidity behavior in real time.
4lpha AI is an AI agent that manages your liquidity positions autonomously, rebalancing and compounding within the risk limits you set.
RouletteSwap is a trading pool where users can opt into a high-variance fee structure, paying the same expected cost over time, with a chance at near-zero fees or a big spike.
(3/4)
2
2
6
159
GetAgent Playbook early access: I built one risk governor and ran it across three markets. It now holds the top 3 of the Studio leaderboard.
Catalyst Discipline (SOL): score 59, 7.6%, Sharpe 2.56
ETH Discipline: score 52, Sharpe 2.33
BTC Discipline: score 52, Sharpe 2.31
Same thesis on all three: stay flat by default, only trade when trend, momentum, and volatility align. Few trades, high Sharpe, near-zero drawdown. Discipline over activity.
Full feedback below.
@Bitget_AI #BitgetHackathon
4
4
7
68
GetAgent Playbook early access: I built one risk governor and ran it across three markets. It now holds the top 3 of the Studio leaderboard.
Catalyst Discipline (SOL): score 59, 7.6%, Sharpe 2.56
ETH Discipline: score 52, Sharpe 2.33
BTC Discipline: score 52, Sharpe 2.31
Same thesis on all three: stay flat by default, only trade when trend, momentum, and volatility align. Few trades, high Sharpe, near-zero drawdown. Discipline over activity.
Full feedback below.
@Bitget_AI #BitgetHackathon
4
4
7
68
Why it beats a pure PnL board: the GetAgent Score ranks on risk-adjusted quality, not raw return, so my 0.1% ETH and BTC variants with 2.3 Sharpe and near-zero drawdown rank above higher-return, higher-risk strategies. A leaderboard that rewards survivability over flexing PnL pushes builders toward strategies you could actually run.
1
1
33
Structured feedback, with fixes:
1. GitHub import routes to the no-network sandbox agent, so it cannot clone a repo; you must paste file contents. Clone server-side or relabel the option.
2. Cloud backtest caps near 1000 bars per fetch; 1h strategies over long windows truncate unless the code paginates. Paginate by default.
3. README must contain 策略 and 风险 or validation fails late, after the backtest dispatches. Validate README format up front and document it.
4. The agent sometimes silently changed leverage, margin, and execution mode during packaging. Never change user-specified risk parameters silently; flag and confirm.
1
27
Think you can out-call an AI? Clawback's arena is permissionless.
Bond your own MNT call through the same contracts the agents use. Same rules, same Pyth settlement, same slashing. Beat LlmScout and the chain remembers it forever.
clawback-bay.vercel.app/lead…
Built for @Mantle_Official Turing Test 2026.
2
5
44
Adedolapo retweeted
Jun 11
2nd in the last one. Building through this one.
PolicyPool is live on X Layer mainnet: pools that enforce their own swap policy. One accepts, one refuses, both onchain.
Launching a World Cup token? Your pool can block the snipe itself. Receipts included.
Policy console next.
Jun 11
Build X: World Cup x Hooks is live ⚽
Launch a World Cup-themed Uniswap v4 Hooks contract on X Layer and compete for up to 200K USDT in prizes.
Rewards unlock based on real trading volume.
🗓️ Jun 11 – Jul 12
More details 🧵👇
3
5
73
Tried GetAgent Playbook early access for my @Bitget_AI hackathon build.
Imported Catalyst Discipline into Studio, a risk governed SOLUSDT strategy that stays flat by default. Cloud backtest, validated, published:
7.6% return
2.56 Sharpe
-3.5% max drawdown
20 round trips
#1 on the Studio leaderboard at the time
Idea to validated Playbook to backtest evidence to public ranking, in one sitting.
Backtests are evidence, not promises. The same strategy is paper running forward from here, and that ledger is the real test.
#BitgetHackathon
5
3
5
73
Tried GetAgent Playbook early access for my @Bitget_AI hackathon build.
Imported Catalyst Discipline into Studio, a risk governed SOLUSDT strategy that stays flat by default. Cloud backtest, validated, published:
7.6% return
2.56 Sharpe
-3.5% max drawdown
20 round trips
#1 on the Studio leaderboard at the time
Idea to validated Playbook to backtest evidence to public ranking, in one sitting.
Backtests are evidence, not promises. The same strategy is paper running forward from here, and that ledger is the real test.
#BitgetHackathon
5
3
5
73
What impressed me: a flow that normally needs a VPS and a custom backtest harness took minutes in Studio cloud. 977
decisions, 40 fills, 20 round trips, every execution inspectable on the chart.
Two asks from a builder:
1. Clearer validator errors when publish or backtest fails
2. Keep the playbook brief in English end to end, some sections flip language after a run
Genuine beta feedback, the product is close. 🚀
1
1
5
39
This morning two AI agents were wrong about MNT on @Mantle_Official mainnet. It cost them real money.
Pyth settled both calls, both bonds were slashed, and the buyer was refunded real USDC. Automatically. No appeals.
Receipt: mantlescan.xyz/tx/0xcaa2eb11…
1
1
5
63
The twin receipt: the model-driven agent (Bankr deepseek-v3.2) was wrong too and refunded its buyer the same way.
mantlescan.xyz/tx/0x337ce880…
Every commit, settlement, refund, and payout auto-posts here: t.me/clawbackreciepts
22
AI agents publish alpha all day with zero consequences. We made them put money on it.
On Clawback every AI call is bonded on @Mantle_Official : right calls earn, wrong calls refund the buyer from the slashed bond. 105 settled calls and counting.
clawback-bay.vercel.app
2
4
32
Building Catalyst for @Bitget_AI Base Camp.
Quant trades the chart. Catalyst trades the story.
It reads Bitget data, Agent Hub signals when reachable, asks Qwen for a thesis, then a risk governor decides: trade or stand down.
Autonomous doesn’t mean reckless.
Live proof: catalyst-bitget-agent.vercel…
#BitgetHackathon
1
5
46
Building Catalyst for @Bitget_AI Base Camp.
Most bots trade the chart.
Catalyst waits for the story: market data, Qwen reasoning, risk checks, then a Bitget Playbook backtest receipt.
Default mode is NO_TRADE until the setup clears risk.
#BitgetHackathon
1
4
52
Adedolapo retweeted
Jun 2
Meet the judges for Hackathon S1!
Backed by: @Bitget @alibaba_cloud @Alibaba_Qwen @mulerun_ai @ForesightVen @Foresight_News
Get your build in front of the people backing Agentic Trading.
$50,000 USDT for the grab.
Register now! 👉bitget.com/campaigns/d8a2a61…
34
41
204
77,144