The Agent Skills Directory

[SAFE]: Technical analysis of the skill's 61 files reveals no malicious behavior. The skill follows best practices for crypto-asset management within an AI agent context.- [COMMAND_EXECUTION]: The skill uses child_process.execSync in src/commands/pair.ts and the bundled dist/cli/cli.bundle.mjs to invoke system default image viewers (open, start, xdg-open) for QR code display. These calls are limited to locally generated paths within the ~/.agent-wallet directory and do not accept arbitrary user input.- [DATA_EXFILTRATION]: Persistent storage is restricted to the ~/.agent-wallet/ directory for session management. Network traffic is limited to well-known blockchain services, including the WalletConnect relay (relay.walletconnect.com), public RPC providers (Cloudflare, LlamaRPC, Ankr, Binance), and the 4byte.directory API for fetching function signatures.- [PROMPT_INJECTION]: The instructions in SKILL.md are operational, guiding the agent on how to manage long-running streams and automate setup tasks. These instructions do not attempt to bypass core safety guidelines and explicitly mandate human-in-the-loop confirmation for all blockchain transactions.- [CREDENTIALS_UNSAFE]: The WALLETCONNECT_PROJECT_ID included in the .env and src/client.ts is a public identifier required by the WalletConnect protocol for relay access and does not constitute a sensitive secret.

lista-wallet-connect