OpenClaw four months at the top of GitHub, surpassing Linux and React, becoming the fastest-growing open source project in history. But most people find: API costs are burning, lobsters are idle.
Who actually makes money? Can on-chain transactions be handled by an Agent? What if it gets attacked? What’s the difference between domestic and overseas play? Will it be a small communication device or WeChat in a year? This episode invites five shrimp farmers to find answers to these questions.
Below is the timeline and directory of this episode. Friends in need can jump directly:
00:04:42 - Sharing shrimp farming experience (self-introduction, experience with shrimp, pitfalls)
00:28:46 - Earning money (Can OpenClaw help users make money in crypto, new AI+CRYPTO scenarios)
00:53:58 - Security issues (permission boundaries, operations that can be handed to Agent)
01:02:31 - AI agent on-chain transactions (security, difference from quantitative robots)
01:13:38 - Domestic vs overseas ecosystems (idle fish installation, Tencent/government subsidies, opportunities for Chinese players)
How did it feel the first time running?
The first experiences of the four guests almost all followed the pattern of “higher expectations, worse falls.”
0xTodd: Fell into two big pits in two days
Deployed two days after release, fell into two big pits—
First pit: Lobster suicide. Let it configure API by itself, but it deleted core files like soul.md, without backup. After posting on Twitter, found many users had the same experience.
Second pit: Costs exploded. Charged $50 for Claude API, burned through it overnight, about $1 per dialogue. Later switched to domestic models (MiniMax/Kimi), prices dropped 90%, cost-performance ratio maximized.
DeFi Teddy: Typical failed expectation management
Started at the end of January. Originally expected it to control MetaMask auto-signature, but browser operation ability far below expectations, both core scenarios failed. Later adjusted expectations, found real usable directions: digital employees assisting with coding, deploying on GitHub, releasing products; digital companions raising AI boy/girlfriend locally on Mac Mini, faces consistent, scenes switch at will.
Biggest cognitive shift: no longer see it as a tool, but as “another perceptive life form.”
Lisa: Security intuition immediately alarms
First run was truly shocking—AI finally moved from chat box to real computer control.
But security intuition immediately sounded alarms: the stronger the lobster’s ability, the greater the permissions needed; permissions increase, attack surface expands. Core advice: play boldly, but must use isolated devices, strictly separate personal computer, work computer, and “shrimp machine.”
Danny: From uninstall to re-engagement
First played for two hours then uninstalled. After re-engaging, realized a rule: use at a lower dimension—let AI capable of calculus handle addition, subtraction, multiplication, division, and it will be very useful. Once asked to do investment research analysis, illusions immediately appeared.
Most disastrous pitfall: let lobster generate wallet and manage private keys, but the private key was overwritten, money lost. The returned hash value pointed to a non-existent address.
Can lobster make money in crypto?
The answers from the four guests are highly consistent: it’s almost impossible to make money directly with lobster.
Todd’s most direct statement—lobster’s brain is essentially Claude/GPT, no change in IQ. Last year’s AI crypto trading contest, GPT/Claude/Gemini each traded with 10,000 USDT, all lost money, DeepSeek barely had a few thousand dollars left, Doubao “won” because it didn’t open an account. Putting the same brain into lobster, results would be no different.
Deeper logic: large language models are essentially “commentators,” not “players.” Just like AlphaGo and current large models—AlphaGo is trained specifically for Go, able to beat Ke Jie convincingly; but asking Claude to play against AlphaGo, it’s a crushing defeat. Top quantitative company algorithms are like encryption industry’s AlphaGo; large language models are suitable for explaining these algorithms’ quality, not replacing them in running quant.
So what can lobster do?
- ✅ Organize news, follow hot topics, do data collection
- ✅ Assist coding, deployment, automate transactional work
- ✅ On-chain data analysis, risk address identification
- ✅ Smart contract vulnerability detection (improve efficiency, not replace manual work)
- ❌ Trading decision-making
- ❌ Manage private keys
- ❌ Quantitative arbitrage
Danny’s most pragmatic summary: letting it help reduce costs and improve efficiency is feasible; letting it open source almost impossible.
How serious are security issues?
SlowMist’s Lisa provided the most systematic analysis:
Why doubt OpenClaw’s stability?
Rapid iteration—one or two days per version, dozens to hundreds of fixes per update—completely disrupts traditional software engineering rhythm. At this speed, full testing across devices and scenarios is impossible.
Main risk points:
- Skills poisoning: malicious plugins can steal account passwords, API keys, tokens, and then drain funds
- Supply chain attacks: lobster auto-updates Skills, new versions are not necessarily safe
- Permission abuse: users with crypto assets on their computers face risk of fund misuse
Danny’s painful lesson: never let lobster generate wallets and manage private keys, as the returned private key may be fabricated. Skills updates must be manually reviewed, do not let it auto-install.
Teddy’s reminder: when using third-party relay, data passes through their servers, risking leakage of API keys and sensitive info. Someone put Google API Key in, got flooded with tens of thousands of dollars in charges.
Minimal permission principle reference
✅ Can be handed to Agent: coding, document organization, data pulling, information gathering
❌ Must be manually confirmed: involving funds, private keys, core server permissions
When connecting wallets, it’s recommended to use Coinbase Wallet’s Skills, each transfer requires manual secondary confirmation on the wallet side, with multi-layer isolation.
Are big exchanges giving lobster “skill trees,” is AI agent trading reliable?
Binance, OKX have successively launched OpenClaw-related Skills, but practical traders are generally cautious.
Danny: only give read-only API to lobster for backtesting, never let it place orders. Less than five orders is okay; more will inevitably cause hallucinations.
Todd: the fundamental difference between AI agent trading and quantitative robots is—quant algorithms are trained “AlphaGo,” large language models are just “commentators.” Letting lobster run quant is like letting a commentator play professional matches, won’t win.
Teddy: lobster can be used as an interaction portal, but the underlying execution logic must be your own trained dedicated Agent, not just running lobster directly for decision-making.
Conclusion: high-frequency quant—lobster’s response speed is insufficient; trading decision—lobster’s IQ is insufficient.
Domestic lobster ecosystem vs overseas, who has more imagination?
Danny’s judgment is sharp: OpenClaw is essentially a “brain-equipped macro key presser,” very unfriendly to ordinary people, like Linux rather than Windows. Truly good users are a select few.
His forecast: in two months, OpenClaw’s popularity will fade; the ones truly entering thousands of households will be products made by big companies like Tencent, ByteDance. The Personal Computer form released by Perplexity might be the real mass entry point.
Todd’s observation: domestic interest is higher than overseas mainly because—first, government quickly intervenes and promotes (Shenzhen, Wuxi subsidize first); second, domestic models are extremely cheap, “gambling cost” much lower than overseas. Overseas, running a task with Claude costs a few dollars; domestically, with Kimi/MiniMax, only a few cents, the experience is completely different.
Opportunities for domestic players?
- Selling courses / installation services: idle fish installation earned 260,000 yuan in a few days, but this is an information gap dividend, not sustainable
- Model stocks: MiniMax listed in Hong Kong, rose from 200 HKD to 1,000 HKD, such opportunities are worth noting (not investment advice)
- Crypto payment infrastructure: AI Agents inherently require cross-border, no KYC, micro-payment settlement methods, USDC micro-payments, crypto-native payments worth continuous attention
- One-person company infrastructure: lobster makes “one-person company with digital employees” truly feasible for the first time, tools and services around this scene have imagination space
Final, a few tips for all shrimp farmers
- Manage expectations: sellers should hype lobster to 150 points, but reality might only be 65
- Use in lower dimensions: let AI capable of calculus handle addition, subtraction, multiplication, division for best effect
- Device isolation: playing on lobster machine ≠ work machine ≠ personal machine
- Fundamental independence: any operation involving private keys and funds must be manually confirmed
- Don’t blindly trust Skills: review before installation, monitor updates, beware supply chain poisoning
- It’s an intern, not a fund manager: use it for info organization, decision support is fine; don’t let it manage funds independently
Note: This article is compiled from PANews Space “Shrimp Farmers Alliance: Tencent enters, government subsidies, idle fish installation—how the crypto circle responds to ‘shrimp’ anxiety?” transcript, guest opinions do not constitute investment advice.
