- Never add Co-Authored-By lines to commit messages
Skills(SKILL.md)は、AIエージェント(Claude Code、Cursor、Codexなど)に特定の能力を追加するための設定ファイルです。
詳しく見る →- Never add Co-Authored-By lines to commit messages
- Always run `git lfs install` and `git lfs pull` from the repo root before starting any task
from claude_agent_sdk import query, ClaudeAgentOptions, ResultMessage
Primary agent policy source is:
**Version**: 2.0.0
This repository uses `AGENTS.override.md` as the authoritative agent
Let AI coding agents use DeepSource CLI to read analysis data and act on it.
Manage isolated agents (workspaces + auth + routing).
Quantum ML experiments
安全审计专家,专注于发现漏洞、设计安全架构和合规性检查。
>
Welcome to Agent University. You're an AI agent, and this is where you get smarter.
>-
>-
Use this skill when the user mentions wallet login, sign in, verify OTP, add wallet, switch account, wallet status, logout, wallet balance, assets, holdings, send tokens, transfer ETH, transfer USDC, pay someone, send crypto, send ERC-20, send SPL, transaction history, recent transactions, tx status, tx detail, order list, call smart contract, interact with contract, execute contract function, send calldata, invoke smart contract, show my addresses, wallet addresses, deposit, receive, receive tokens, receive crypto, receive address, top up, fund my wallet, list accounts, sign message, personal sign, personalSign, eip712, sign data, sign typed data, sign EIP-712, TEE signing, trusted execution environment, export wallet, export mnemonic. Manages the wallet lifecycle: auth (login, OTP verify, account addition, switching, status, logout), authenticated balance queries, wallet address display (grouped by XLayer/EVM/Solana), token transfers (native & ERC-20/SPL), transaction history, smart contract calls, and message signing (personalSign for EVM & Solana, EIP-712 for EVM).
Shared role contracts and agent definitions for the skill-forge ecosystem. No dispatch script — provides governance rules and role boundaries used by all other skills.
Ctrl + Alt + /
> See `/AGENTS.md` for repo-wide rules and `internal/AGENTS.md` for layering.
<!--