AKShare

Use AKShare for Chinese market and macro-finance data via Python. Use when the user asks for A股、港股、美股、ETF、基金、指数、宏观、利率、债券、期货、商品、分红、财务 or other public-market d...

This skill does what it says (install AKShare and run queries) but the helper executes whatever Python expression you pass. That means a crafted expression could read local files, environment variables (including secrets), or perform network requests. Before installing or allowing the agent to invoke this skill autonomously: 1) Only use it if you trust the skill author and the environment where the venv will run. 2) Consider running the bootstrap and evaluation commands manually in an isolated sandbox (container or dedicated VM). 3) If you need to allow automated use, request or implement a safer evaluator that restricts builtins and blocks __import__/os/subprocess or only accepts pre-validated AKShare function calls. 4) Do not supply sensitive credentials to the agent when using this skill. If you want, ask the maintainer to replace eval with a safe execution model (e.g., whitelist of AKShare functions or AST-based expression validation).

Type: OpenClaw Skill Name: zack995-akshare Version: 0.1.0 The skill contains a significant security vulnerability in 'scripts/akshare_eval.py', which uses the 'eval()' function to execute arbitrary Python code passed via the '--expr' argument. While this is a common (though risky) pattern for allowing an AI agent to execute dynamic data-fetching commands, it creates a Remote Code Execution (RCE) surface. The 'scripts/bootstrap_akshare_env.sh' script performs standard environment setup, and the overall functionality is consistent with the stated goal of fetching financial data via the AKShare library, but the lack of input sanitization in the execution script warrants a suspicious classification.