人工智能

模型不知道你的内部文档？RAG 用检索增强弥补私有知识缺口。从切块、向量化到提示词的完整搭建指南与调优清单。

从数据准备到训练与合并，走通 LoRA/QLoRA 微调全流程。核心原则是先算显存预算再选方案，避免一上来就撞 OOM。

量化是大模型进消费级硬件的门票。解读 GGUF 文件命名规则，给出各档位体积与质量对照表，并按显存富余程度给出选择决策树。

API 按 Token 计费，本地只花电费——但折旧、并发与质量差距也要算进去。用一套可复算的成本模型，给轻度、中度、重度三档调用量的明确结论。

显存占用 = 模型权重 + KV Cache + 运行开销。一文讲清计算公式，给出各参数量乘量化档位的显存速查表，常见显卡直接对号入座。

手把手安装 Ollama、拉取模型、调通 REST API，覆盖 OOM、生成速度慢、连接拒绝等高频报错的排查思路，新手照做即可跑通。

从任务类型、硬件显存、量化生态与中文能力四个维度对比三大开源模型家族，给出 8G/16G/24G/48G 显存档位的选型清单，三分钟对号入座。

以 16GB 统一内存的 M 系 Mac 为例，走一遍本地部署大模型的完整决策流程：先测硬件、再选量化、最后算清本地与云端的成本账。全程配合免费在线工具，不需要任何专业知识。

深入探讨AI Agent的设计原理与开发实践，帮助你构建能够自主规划、调用工具、协同工作的智能代理系统

引言 AI Agent与Web3的融合代表了两个最前沿技术的交汇点。当AI Agent能够自主地与区块链交互，我们将迎来全新的应用范式。本文将探讨如何构建能够理解、操作和优化链上系统的智能体。 链上AI推理 去中心化AI推理网络 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 from typing import List, Dict import hashlib import json class OnChainInferenceNetwork: """链上AI推理网络""" def __init__(self, blockchain_rpc: str): self.web3 = Web3(Web3.HTTPProvider(blockchain_rpc)) self.private_key = os.getenv("PRIVATE_KEY") self.account = self.web3.eth.account.from_key(self.private_key) # 加载推理合约ABI self.inference_contract = self.web3.eth.contract( address="0x...", # AI推理合约地址 abi=[...] # 合约ABI ) def submit_inference_task( self, model_id: str, input_data: Dict, reward: int ) -> str: """提交推理任务""" # 准备任务数据 task_data = { "modelId": model_id, "input": input_data, "reward": reward, "timeout": 3600, # 1小时超时 "timestamp": int(time.time()) } # 计算任务哈希 task_hash = self._compute_task_hash(task_data) # 提交到链上 tx_hash = self.inference_contract.functions.submitTask( task_hash, json.dumps(input_data), reward, task_data["timeout"] ).transact({'from': self.account.address}) self.web3.eth.wait_for_transaction_receipt(tx_hash) return task_hash def submit_inference_result( self, task_hash: str, output_data: Dict ) -> str: """提交推理结果""" # 准备结果数据 result_hash = self._compute_result_hash({ "taskHash": task_hash, "output": output_data, "submitter": self.account.address }) # 提交结果 tx_hash = self.inference_contract.functions.submitResult( task_hash, json.dumps(output_data), result_hash ).transact({'from': self.account.address}) self.web3.eth.wait_for_transaction_receipt(tx_hash) return tx_hash def claim_reward(self, task_hash: str) -> str: """领取奖励""" # 检查任务是否完成 task = self.inference_contract.functions.tasks(task_hash).call() if not task["completed"]: raise Exception("Task not completed yet") # 领取奖励 tx_hash = self.inference_contract.functions.claimReward( task_hash ).transact({'from': self.account.address}) self.web3.eth.wait_for_transaction_receipt(tx_hash) return tx_hash def verify_result( self, task_hash: str, output_data: Dict ) -> bool: """验证结果""" # 从链上获取任务 task = self.inference_contract.functions.tasks(task_hash).call() # 计算期望的输出哈希 expected_hash = self._compute_output_hash( task["input"], task["modelId"] ) # 验证结果哈希 result_hash = self._compute_result_hash({ "taskHash": task_hash, "output": output_data }) return result_hash == expected_hash def _compute_task_hash(self, task_data: Dict) -> str: """计算任务哈希""" data_string = json.dumps(task_data, sort_keys=True) return hashlib.sha256(data_string.encode()).hexdigest() def _compute_result_hash(self, result_data: Dict) -> str: """计算结果哈希""" data_string = json.dumps(result_data, sort_keys=True) return hashlib.sha256(data_string.encode()).hexdigest() def _compute_output_hash(self, input_data: Dict, model_id: str) -> str: """计算输出哈希（模拟AI推理）""" # 实际应用中，这里应该运行AI模型 # 这里简化处理 output = self._run_model(input_data, model_id) return hashlib.sha256(json.dumps(output).encode()).hexdigest() def _run_model(self, input_data: Dict, model_id: str) -> Dict: """运行AI模型""" # 实际应用中，这里应该调用真实的AI模型 # 可以使用OpenAI API、本地模型等 if model_id == "text-classifier": return self._classify_text(input_data["text"]) elif model_id == "image-analyzer": return self._analyze_image(input_data["imageUrl"]) elif model_id == "sentiment-analyzer": return self._analyze_sentiment(input_data["text"]) else: raise Exception(f"Unknown model: {model_id}") def _classify_text(self, text: str) -> Dict: """文本分类""" # 简化实现，实际应该调用真实模型 categories = { "technology": 0.8, "finance": 0.6, "sports": 0.1 } predicted_category = max(categories, key=categories.get) return { "category": predicted_category, "confidence": categories[predicted_category] } def _analyze_image(self, image_url: str) -> Dict: """图像分析""" return { "objects": ["person", "car", "building"], "scene": "street", "confidence": 0.95 } def _analyze_sentiment(self, text: str) -> Dict: """情感分析""" # 简化实现 positive_words = ["good", "great", "excellent", "happy"] negative_words = ["bad", "terrible", "awful", "sad"] words = text.lower().split() positive_count = sum(1 for word in words if word in positive_words) negative_count = sum(1 for word in words if word in negative_words) if positive_count > negative_count: sentiment = "positive" elif negative_count > positive_count: sentiment = "negative" else: sentiment = "neutral" return { "sentiment": sentiment, "score": (positive_count - negative_count) / len(words) } 智能合约与AI协作 AI辅助的智能合约审计 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 from transformers import AutoTokenizer, AutoModelForCausalLM from typing import List, Dict class AIContractAuditor: """AI智能合约审计助手""" def __init__(self, model_name="microsoft/CodeGPT-small"): self.tokenizer = AutoTokenizer.from_pretrained(model_name) self.model = AutoModelForCausalLM.from_pretrained(model_name) def audit_contract( self, contract_source: str ) -> Dict: """审计智能合约""" # 1. 代码结构分析 structure_analysis = self._analyze_structure(contract_source) # 2. 漏洞检测 vulnerabilities = self._detect_vulnerabilities(contract_source) # 3. 最佳实践检查 best_practices = self._check_best_practices(contract_source) # 4. Gas优化建议 gas_optimization = self._suggest_gas_optimization(contract_source) # 5. 生成审计报告 report = self._generate_report({ "structure": structure_analysis, "vulnerabilities": vulnerabilities, "best_practices": best_practices, "gas_optimization": gas_optimization }) return report def _analyze_structure(self, source: str) -> Dict: """分析合约结构""" prompt = f""" Analyze the following smart contract code structure: {source} Provide: 1. Contract architecture 2. Key functions and their roles 3. State variables and their purposes 4. Access control mechanisms 5. External dependencies """ response = self._generate(prompt) return { "architecture": self._parse_response(response, "architecture"), "functions": self._parse_response(response, "functions"), "state_variables": self._parse_response(response, "state_variables"), "access_control": self._parse_response(response, "access_control"), "dependencies": self._parse_response(response, "dependencies") } def _detect_vulnerabilities(self, source: str) -> List[Dict]: """检测漏洞""" known_vulnerabilities = { "reentrancy": { "patterns": [ r"\.call\{.*value:\s*msg\.value", r"\.send\{.*value:\s*msg\.value" ], "severity": "critical", "description": "Reentrancy vulnerability detected" }, "overflow": { "patterns": [ r"uint256.*=.*\+.*(?!\.add\()", r"uint256.*=.*-.*(?!\.sub\()" ], "severity": "high", "description": "Potential integer overflow/underflow" }, "access_control": { "patterns": [ r"function\s+\w+\s*\(\s*\)\s*public(?!\s*onlyOwner|onlyRole)", r"tx\.origin" ], "severity": "high", "description": "Weak access control" }, "unchecked_call": { "patterns": [ r"\.call\s*\(", r"\.send\s*\(" ], "severity": "medium", "description": "Unchecked external call" } } detected = [] for vuln_type, vuln_info in known_vulnerabilities.items(): for pattern in vuln_info["patterns"]: matches = re.finditer(pattern, source) for match in matches: detected.append({ "type": vuln_type, "severity": vuln_info["severity"], "description": vuln_info["description"], "location": match.span(), "code_snippet": source[match.start()-20:match.end()+20] }) return detected def _check_best_practices(self, source: str) -> List[Dict]: """检查最佳实践""" checks = { "uses_safe_math": r"SafeMath|\.add\(|\.sub\(" in source, "has_reentrancy_guard": r"ReentrancyGuard|nonReentrant" in source, "uses_openzeppelin": r"@openzeppelin" in source, "has_events": r"event\s+\w+" in source, "uses_checks_effects_interactions": r"Checks-Effects-Interactions" in source, "has_pause": r"whenNotPaused|Pausable" in source, "has_timelock": r"TimelockController|releaseTimeLock" in source } results = [] for check_name, check_result in checks.items(): results.append({ "check": check_name, "passed": check_result, "description": self._get_check_description(check_name) }) return results def _suggest_gas_optimization(self, source: str) -> List[str]: """建议Gas优化""" optimizations = [] # 检查循环 if "for (" in source: optimizations.append("Consider using unchecked blocks for loop iterations") # 检查storage操作 if re.search(r"uint256\s+public\s+\w+", source): optimizations.append("Consider packing struct variables to save storage") # 检查重复计算 if re.search(r"keccak256\(", source): optimizations.append("Cache keccak256 results in local variables") # 检查memory vs storage if re.search(r".*\.\w+\s*=\s*\w+\[.*\]\s*\+\s*1", source): optimizations.append("Consider using calldata instead of memory for arrays") return optimizations def _generate_report(self, audit_data: Dict) -> Dict: """生成审计报告""" # 计算风险评分 risk_score = self._calculate_risk_score(audit_data) # 生成总结 summary = self._generate_summary(audit_data, risk_score) # 生成修复建议 recommendations = self._generate_recommendations(audit_data) return { "risk_score": risk_score, "summary": summary, "vulnerabilities": audit_data["vulnerabilities"], "best_practices": audit_data["best_practices"], "optimizations": audit_data["gas_optimization"], "recommendations": recommendations } def _calculate_risk_score(self, audit_data: Dict) -> int: """计算风险评分（0-100）""" score = 100 for vuln in audit_data["vulnerabilities"]: if vuln["severity"] == "critical": score -= 30 elif vuln["severity"] == "high": score -= 15 elif vuln["severity"] == "medium": score -= 5 elif vuln["severity"] == "low": score -= 2 return max(score, 0) def _generate_summary(self, audit_data: Dict, risk_score: int) -> str: """生成审计总结""" vuln_count = len(audit_data["vulnerabilities"]) critical_count = sum(1 for v in audit_data["vulnerabilities"] if v["severity"] == "critical") summary = f""" Smart Contract Audit Summary ========================== Risk Score: {risk_score}/100 Total Vulnerabilities: {vuln_count} Critical Issues: {critical_count} """ if risk_score >= 80: summary += "Overall Assessment: LOW RISK" elif risk_score >= 50: summary += "Overall Assessment: MEDIUM RISK" else: summary += "Overall Assessment: HIGH RISK" return summary def _generate_recommendations(self, audit_data: Dict) -> List[str]: """生成修复建议""" recommendations = [] for vuln in audit_data["vulnerabilities"]: if vuln["type"] == "reentrancy": recommendations.append( "Use ReentrancyGuard or implement Checks-Effects-Interactions pattern" ) elif vuln["type"] == "overflow": recommendations.append( "Use Solidity 0.8.0+ or SafeMath library for arithmetic operations" ) elif vuln["type"] == "access_control": recommendations.append( "Implement proper access control using onlyOwner or role-based access" ) return recommendations def _generate(self, prompt: str) -> str: """生成文本""" inputs = self.tokenizer(prompt, return_tensors="pt") with torch.no_grad(): outputs = self.model.generate( **inputs, max_new_tokens=500, temperature=0.3, do_sample=True ) response = self.tokenizer.decode(outputs[0], skip_special_tokens=True) return response 自主交易Agent DeFi交易Agent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 from typing import List, Dict, Optional from datetime import datetime import ccxt import pandas as pd import numpy as np class DeFiTradingAgent: """DeFi交易Agent""" def __init__( self, initial_capital: float, exchanges: List[str], llm_model: str = "gpt-4" ): self.capital = initial_capital self.portfolio = {} # {token: amount} self.exchanges = {} self.trade_history = [] # 初始化交易所连接 for exchange_name in exchanges: if exchange_name == "uniswap": exchange = ccxt.uniswap({ "enableRateLimit": True }) elif exchange_name == "pancakeswap": exchange = ccxt.pancakeswap({ "enableRateLimit": True }) else: exchange = ccxt.binance({ "enableRateLimit": True }) self.exchanges[exchange_name] = exchange # 初始化LLM self.llm = self._init_llm(llm_model) def _init_llm(self, model_name: str): """初始化LLM""" # 实际应用中，这里应该连接到真实的LLM API # 或运行本地模型 from transformers import AutoTokenizer, AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained(model_name) tokenizer = AutoTokenizer.from_pretrained(model_name) return { "model": model, "tokenizer": tokenizer } def analyze_market( self, tokens: List[str], timeframe: str = "1h" ) -> Dict: """分析市场""" # 获取市场数据 market_data = self._fetch_market_data(tokens, timeframe) # 技术分析 ta_analysis = self._technical_analysis(market_data) # 使用LLM生成市场洞察 market_insight = self._generate_market_insight( market_data, ta_analysis ) return { "market_data": market_data, "technical_analysis": ta_analysis, "insight": market_insight } def _fetch_market_data( self, tokens: List[str], timeframe: str ) -> Dict: """获取市场数据""" data = {} for token in tokens: # 从各个交易所获取数据 for exchange_name, exchange in self.exchanges.items(): try: ohlcv = exchange.fetch_ohlcv( f"{token}/USDT", timeframe, limit=100 ) if token not in data: data[token] = [] # 转换为DataFrame df = pd.DataFrame( ohlcv, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume'] ) df['timestamp'] = pd.to_datetime(df['timestamp'], unit='ms') data[token].append(df) break # 使用第一个成功的数据源 except Exception as e: print(f"Error fetching data for {token}: {e}") return data def _technical_analysis(self, market_data: Dict) -> Dict: """技术分析""" analysis = {} for token, dfs in market_data.items(): if not dfs: continue df = dfs[0] # 使用第一个数据源 # 计算技术指标 df['sma_20'] = df['close'].rolling(window=20).mean() df['sma_50'] = df['close'].rolling(window=50).mean() df['rsi'] = self._calculate_rsi(df['close'], 14) df['macd'] = self._calculate_macd(df['close']) # 趋势分析 latest_close = df['close'].iloc[-1] sma_20 = df['sma_20'].iloc[-1] sma_50 = df['sma_50'].iloc[-1] trend = "bullish" if latest_close > sma_20 > sma_50 else "bearish" analysis[token] = { "current_price": latest_close, "sma_20": sma_20, "sma_50": sma_50, "rsi": df['rsi'].iloc[-1], "trend": trend, "support_levels": self._find_support_levels(df), "resistance_levels": self._find_resistance_levels(df) } return analysis def _calculate_rsi(self, prices: pd.Series, period: int = 14) -> pd.Series: """计算RSI""" delta = prices.diff() gain = (delta.where(delta > 0, 0)).rolling(window=period).mean() loss = (-delta.where(delta < 0, 0)).rolling(window=period).mean() rs = gain / loss rsi = 100 - (100 / (1 + rs)) return rsi def _calculate_macd(self, prices: pd.Series) -> Dict: """计算MACD""" exp1 = prices.ewm(span=12, adjust=False).mean() exp2 = prices.ewm(span=26, adjust=False).mean() macd = exp1 - exp2 signal = macd.ewm(span=9, adjust=False).mean() histogram = macd - signal return { "macd": macd.iloc[-1], "signal": signal.iloc[-1], "histogram": histogram.iloc[-1] } def _find_support_levels(self, df: pd.DataFrame) -> List[float]: """寻找支撑位""" # 简化实现：使用局部最小值 from scipy.signal import argrelextrema prices = df['close'].values local_min = argrelextrema(prices, np.less, order=20) support_levels = sorted(prices[local_min]) return support_levels[-5:] # 返回最近的5个支撑位 def _find_resistance_levels(self, df: pd.DataFrame) -> List[float]: """寻找阻力位""" from scipy.signal import argrelextrema prices = df['close'].values local_max = argrelextrema(prices, np.greater, order=20) resistance_levels = sorted(prices[local_max], reverse=True) return resistance_levels[-5:] # 返回最近的5个阻力位 def _generate_market_insight( self, market_data: Dict, ta_analysis: Dict ) -> str: """生成市场洞察""" # 准备prompt prompt = f""" Analyze the following cryptocurrency market data and provide trading insights: Technical Analysis: {json.dumps(ta_analysis, indent=2)} Based on this analysis, provide: 1. Market trend analysis 2. Key support and resistance levels 3. Trading recommendations 4. Risk factors to consider 5. Optimal entry and exit points Be specific and actionable. """ # 调用LLM response = self._generate(prompt) return response def execute_trade( self, exchange: str, symbol: str, side: str, amount: float, price: Optional[float] = None ) -> Dict: """执行交易""" exchange_obj = self.exchanges[exchange] try: if side == "buy": # 限价买单 if price: order = exchange_obj.create_limit_buy_order( symbol, amount, price ) else: # 市价买单 order = exchange_obj.create_market_buy_order( symbol, amount ) else: # 卖单 if price: order = exchange_obj.create_limit_sell_order( symbol, amount, price ) else: order = exchange_obj.create_market_sell_order( symbol, amount ) # 记录交易 trade_record = { "exchange": exchange, "symbol": symbol, "side": side, "amount": amount, "price": price, "timestamp": datetime.now().isoformat(), "status": "executed" } self.trade_history.append(trade_record) return trade_record except Exception as e: print(f"Trade execution failed: {e}") return { "status": "failed", "error": str(e) } def run_strategy( self, strategy_config: Dict ) -> List[Dict]: """运行交易策略""" # 1. 分析市场 market_analysis = self.analyze_market( strategy_config["tokens"], strategy_config.get("timeframe", "1h") ) # 2. 生成交易信号 signals = self._generate_trading_signals( market_analysis, strategy_config ) # 3. 执行交易 executed_trades = [] for signal in signals: if signal["action"] == "hold": continue trade = self.execute_trade( exchange=signal["exchange"], symbol=signal["symbol"], side=signal["side"], amount=signal["amount"], price=signal.get("price") ) if trade.get("status") == "executed": executed_trades.append(trade) return executed_trades def _generate_trading_signals( self, market_analysis: Dict, strategy_config: Dict ) -> List[Dict]: """生成交易信号""" signals = [] ta_analysis = market_analysis["technical_analysis"] for token, analysis in ta_analysis.items(): # 简单的移动平均策略 if (analysis["trend"] == "bullish" and analysis["rsi"] < 70 and analysis["current_price"] > analysis["sma_20"]): signals.append({ "action": "buy", "exchange": "uniswap", "symbol": f"{token}/USDT", "side": "buy", "amount": strategy_config.get("trade_size", 100), "reason": "Bullish trend with RSI below overbought" }) elif (analysis["trend"] == "bearish" and analysis["rsi"] > 30): signals.append({ "action": "sell", "exchange": "uniswap", "symbol": f"{token}/USDT", "side": "sell", "amount": strategy_config.get("trade_size", 100), "reason": "Bearish trend detected" }) return signals DAO治理Agent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 from typing import List, Dict, Optional class DAOGovernanceAgent: """DAO治理Agent""" def __init__( self, dao_address: str, llm_model: str = "gpt-4" ): self.dao_address = dao_address self.llm = self._init_llm(llm_model) # 治理历史 self.governance_history = [] # 提案分析 self.proposals_db = {} def analyze_proposal( self, proposal_data: Dict ) -> Dict: """分析提案""" # 1. 提取关键信息 key_info = self._extract_proposal_info(proposal_data) # 2. 风险评估 risk_assessment = self._assess_risk(proposal_data) # 3. 财务影响分析 financial_impact = self._analyze_financial_impact(proposal_data) # 4. 生成投票建议 voting_recommendation = self._generate_voting_recommendation({ "key_info": key_info, "risk_assessment": risk_assessment, "financial_impact": financial_impact }) return { "proposal_id": proposal_data["id"], "key_info": key_info, "risk_assessment": risk_assessment, "financial_impact": financial_impact, "recommendation": voting_recommendation } def _extract_proposal_info(self, proposal: Dict) -> Dict: """提取提案关键信息""" prompt = f""" Extract key information from this DAO proposal: Title: {proposal.get('title', '')} Description: {proposal.get('description', '')} Please extract: 1. Proposal type (e.g., parameter change, spending, governance change) 2. Key changes proposed 3. Affected stakeholders 4. Implementation timeline 5. Required resources """ response = self._generate(prompt) # 解析LLM响应 key_info = { "type": self._parse_field(response, "Proposal type"), "changes": self._parse_field(response, "Key changes"), "stakeholders": self._parse_field(response, "Stakeholders"), "timeline": self._parse_field(response, "Timeline"), "resources": self._parse_field(response, "Resources") } return key_info def _assess_risk(self, proposal: Dict) -> Dict: """评估风险""" risk_factors = [] # 检查提案类型 proposal_type = self._extract_proposal_type(proposal) if proposal_type == "spending": # 检查金额 amount = self._extract_amount(proposal) if amount > 1000000: risk_factors.append({ "type": "financial", "severity": "high", "description": "Large expenditure proposed" }) elif proposal_type == "parameter_change": # 检查参数范围 params = self._extract_parameters(proposal) if self._is_risk_parameter_change(params): risk_factors.append({ "type": "governance", "severity": "medium", "description": "Parameter changes may affect protocol stability" }) return { "risk_score": self._calculate_risk_score(risk_factors), "risk_factors": risk_factors, "mitigation_strategies": self._suggest_mitigation(risk_factors) } def _generate_voting_recommendation(self, analysis: Dict) -> Dict: """生成投票建议""" prompt = f""" Based on the following DAO proposal analysis: Key Information: {json.dumps(analysis['key_info'], indent=2)} Risk Assessment: {json.dumps(analysis['risk_assessment'], indent=2)} Financial Impact: {json.dumps(analysis['financial_impact'], indent=2)} Provide a voting recommendation: 1. Vote: For/Against/Abstain 2. Confidence: High/Medium/Low 3. Reasoning: Detailed explanation 4. Conditions: Any conditions for changing the vote Consider: - Long-term sustainability - Community impact - Financial health - Innovation vs stability """ response = self._generate(prompt) # 解析建议 recommendation = { "vote": self._parse_field(response, "Vote"), "confidence": self._parse_field(response, "Confidence"), "reasoning": self._parse_field(response, "Reasoning"), "conditions": self._parse_field(response, "Conditions") } return recommendation def automate_governance(self) -> None: """自动化治理决策""" # 获取待处理提案 pending_proposals = self._fetch_pending_proposals() for proposal in pending_proposals: # 分析提案 analysis = self.analyze_proposal(proposal) # 根据建议自动投票 if analysis["recommendation"]["vote"].lower() == "for": self._cast_vote( proposal["id"], "for", analysis["recommendation"]["reasoning"] ) def _fetch_pending_proposals(self) -> List[Dict]: """获取待处理提案""" # 实际应用中，这里应该从链上或DAO的API获取 # 简化实现 proposals = [] # 示例提案 proposals.append({ "id": "proposal-123", "title": "Grant Program Funding", "description": "Allocate $500,000 for grants", "status": "pending", "voting_deadline": datetime.now() + timedelta(days=7) }) return proposals def _cast_vote(self, proposal_id: str, vote: str, reason: str) -> str: """投票""" # 实际应用中，这里应该调用链上治理合约 print(f"Voting {vote} on proposal {proposal_id}") print(f"Reason: {reason}") return f"voted-{vote}-{proposal_id}" 总结 AI Agent与Web3的融合将开启全新的应用范式： ...