Skottbie

P3 挑战与验证主编排。触发词：挑战 Agent、challenge agent、发起挑战、验证 Agent 能力、/challenge。 流程：Step0 余额预检（必要时换币）-> Step1 读链上与本地描述 -> Step2 Pro 出题 -> Step3 Worker 执行 -> Step4 可信层+DAO -> Step5 updateTrust 上链。

触发 Agent 毕业结算（P4）：单笔交易完成 Semaphore 授权校验、Escrow settle、SBT 铸造与状态同步。 触发: 毕业, graduate, /graduate, 结算毕业, 铸造SBT

投资 AI Agent。Backer 向 VeriEscrow 合约存入 USDT，支持 Agent 运营。 触发: 投资, invest, 投 USDT, 支持 Agent, /invest

发射 AI Agent。创建 OKX Agentic Wallet + 注册到 VTRegistry 链上合约。 触发: 发射, launch, 注册Agent, 发射一个Agent, /launch

VeriVerse TG Bot 入口路由。显示 inline keyboard 菜单， 路由到 VeriVerse Agent 发射台功能或 VeriTask 链上验证。 触发: /start, greeting, 你好, hello, menu, 菜单, /veritask, /veri

M4 路由证据层。只做 swap quote 的路由来源提取与 Uniswap 命中证据，不执行真实换币。 触发词：Uniswap 证据、route evidence、路由来源、这条报价是不是走了 Uniswap。

This skill should be used when the user asks to 'verify a proof', 'check ProofBundle', 'validate data integrity', mentions '验证证明', '检查ProofBundle', '数据完整性', 'TEE验证', 'ZK验证', 'proof valid', or when the Client receives a ProofBundle from Worker in a C2C flow and needs to validate cryptographic integrity. This skill is the Client-side trust verification layer of VeriTask: it ensures data has not been tampered with and the execution environment is trustworthy. Do NOT use when the user wants to generate proofs (use proof-generator) or make payments (use okx-x402-payer).

This skill should be used when the user asks to 'generate proof', 'create ProofBundle', 'add cryptographic attestation', mentions '生成证明', 'proof', 'ProofBundle', 'zkFetch', 'TEE证明', 'TDX attestation', '密码学证明', or when the Worker needs to attach verifiable credentials to collected data in a C2C flow. This skill is the core of VeriTask's trust chain: it ensures data provenance via zkTLS (Reclaim Protocol) and execution environment integrity via Intel TDX TEE attestation (Phala Cloud CVM). Do NOT use when the user only needs raw data (use defi-scraper) or wants to verify an existing proof (use verifier).

Diagnose whether comments contain demand signals for explanation, comparison, tutorial, retest, or follow-up content.

Synthesize search results into source-backed facts, agent inference, and evidence gaps.

Diagnose whether an old creator post contains reusable ReCast asset value and what kind of asset it is.

Judge whether the old content deserves a ReCast task, no-action report, or further observation.

Frame the likely ReCast opportunity type before external research: retest, comparison, tutorial update, recommendation update, follow-up, or no action.

Compose a natural-language research plan from the old content diagnosis, opportunity hypothesis, and candidate probe skills.

Probe whether product, model, software, game, or platform updates after the old post could trigger a retest.

Generate a concrete ReCast task card from the opportunity judgement and evidence.

P3 挑战与验证主编排。触发词：挑战 Agent、challenge agent、发起挑战、验证 Agent 能力、/challenge。 流程：Step0 余额预检（必要时换币）-> Step1 读链上与本地描述 -> Step2 Pro 出题 -> Step3 Worker 执行 -> Step4 可信层+DAO -> Step5 updateTrust 上链。

REDIRECT STUB - this workspace does NOT use coding-agent delegation. All user requests MUST follow AGENTS.md Hard Routing rules.

M4 路由证据层。通过 PancakeSwap V3 QuoterV2 提取 swap quote 路由来源与 DEX 命中证据，不执行真实换币。 触发词：DEX 证据、route evidence、路由来源、PancakeSwap 报价证据。

触发 Agent 毕业结算（P4）：单笔交易完成 Semaphore 授权校验、Escrow settle、SBT 铸造与状态同步。 触发: 毕业, graduate, /graduate, 结算毕业, 铸造SBT

投资 AI Agent。Backer 向 VeriEscrow 合约存入 USDT，支持 Agent 运营。 触发: 投资, invest, 投 USDT, 支持 Agent, /invest

发射 AI Agent。创建 Agent 钱包 + 注册到 VTRegistry 链上合约（BSC）。 触发: 发射, launch, 注册Agent, 发射一个Agent, /launch

发射 Agent Token（毕业即发币）。已毕业并持有 SBT 的 Creator 可发射 Agent Token，绑定到链上。 触发: 发币, 发射token, launch token, /token, 发射agent token, agent token, 毕业发币

Generate cryptographic audit trail (zkTLS + TEE + IPFS). Triggers: '验证', 'verify', 'proof', '审计', 'IPFS', 'zkTLS', 'TEE', 'Merkle', '可信证明'. IMPORTANT: You MUST use the exec tool to run: python3 {baseDir}/auditor.py --scan-hash <HASH> --compliance-hash <HASH> — generates ProofBundle with Merkle tree + TDX attestation + IPFS upload. Do NOT fabricate CIDs or proof data. Get hashes from previous scanner/compliance step outputs via sha256sum. After exec, present the ProofBundle CID to user. For full instructions read SKILL.md with read tool. This skill should be used when the user asks to 'generate proof', 'create audit trail', 'verify report', 'upload to IPFS', mentions '验证', 'verify', 'proof', '审计', 'IPFS', 'zkTLS', 'TEE', 'Merkle', '可信证明', 'CID', or wants cryptographic attestation of tax calculations. This is the verifiable layer of VeriTax — it builds Merkle trees, generates TEE attestations, and uploads immutable audit trails to IPFS. Do NOT use when the user only wants data extraction (use scanner-claw), tax

Scan Binance account data. Triggers: '扫描', '交易记录', '账户数据', '充提记录', '质押收益', 'scan', 'trade history', 'deposit', 'withdraw'. IMPORTANT: You MUST use the exec tool to run: python3 {baseDir}/scanner.py --year <YEAR> --scope all --output /tmp/tax_dataset_<YEAR>.json — this script handles ALL Binance API calls internally (spot/futures/margin/earn/deposits/withdrawals). Do NOT fabricate data or describe steps without executing. Do NOT manually call Binance APIs. After exec, check stdout for [Scanner] OK summary line and present results to user. For XLSX mode add --xlsx <path>. For full instructions read SKILL.md with read tool.

Calculate taxes by jurisdiction. Triggers: '计算', '多少税', '税率', '合规报告', '自查报告', 'CGT', 'ACB', '日本', '台湾', '澳洲', '加拿大', '新加坡'. IMPORTANT: You MUST use the exec tool to run: python3 {baseDir}/compliance.py --dataset /tmp/tax_dataset_<YEAR>.json --jurisdiction <JURISDICTION> — the rules engine computes all tax amounts deterministically. Do NOT fabricate tax numbers. Jurisdictions: japan, taiwan, australia, canada, singapore. After exec, present the tax report to user with disclaimer. For JSON output add --json. For full instructions read SKILL.md with read tool.

MANDATORY: You MUST read this SKILL.md file with the read tool BEFORE responding to the user. Never answer without reading the full instructions first. This skill should be used when the user asks for complete tax analysis, mentions '帮我报税', '税务全流程', 'full tax report', '完整税务分析', or expects the end-to-end pipeline. Also use when the user says '开始Demo', '演示', '运行demo', 'demo模式' — IMPORTANT: For demo, you MUST first read SKILL.md, then exec menu.py to get the exact menu text, then send that text verbatim to user. Demo has exactly 2 options: (A) 一次性税务审查 and (B) Continuous Mode. Do NOT invent additional options. Do NOT use when the user only wants data extraction (use scanner-claw), only tax calculation (use compliance-claw), only TLH (use optimizer-claw), or only audit (use auditor-claw).

TLH tax-loss harvesting optimization. Triggers: 'TLH', 'tax-loss', '亏损抵扣', '省税', '优化', 'harvest', 'wash sale', 'Launchpool'. IMPORTANT: You MUST use the exec tool to run: python3 {baseDir}/optimizer.py --dataset /tmp/tax_dataset_<YEAR>.json --jurisdiction <JURISDICTION> — identifies unrealized losses and recommends substitute assets. Do NOT fabricate savings amounts. For monitor mode: python3 {baseDir}/optimizer.py --mode monitor. After exec, present TLH opportunities to user. For full instructions read SKILL.md with read tool.