SQL Database Toolkit

All-in-one SQL data analysis toolkit supporting database/file connection, SQL query, visualization, AI insights, and report/dashboard generation with templates.

What to consider before installing/using this skill: - Code vs metadata mismatch: the package includes many Python scripts and an install_deps.sh but the registry declares no install spec. Do not blindly run install_deps.sh or pip install -r requirements.txt; inspect those files first. - Credentials: this toolkit needs DB credentials to function. Do not provide production credentials. Prefer creating a dedicated, least-privilege, readonly database user or use local/test databases first. The SKILL.md references ~/.my.cnf — check any code that reads local config files before use. - Network access: interactive charts use CDNs (e.g., jsdelivr) and the code may perform network I/O; review code for external endpoints (search for 'requests', 'urllib', 'socket', 'http', 'https', or hardcoded URLs) before running. - Run in isolation: if you want to evaluate, run the code in a disposable environment (Python venv or container) and monitor outbound network traffic. Review requirements.txt and dependency versions for supply-chain risk. - Ask the publisher for clarifications: request a homepage or repository link, an explicit install spec, and explicit declarations of required environment variables (e.g., DB credentials) and any network endpoints the skill contacts. Also ask them to explain why metadata lists version 1.0.0 while SKILL.md says v2.0.0. - If you lack the ability to audit code, avoid running it with sensitive credentials. Use sample data or a sandboxed DB and inspect outputs/logs first. If the publisher provides a public repository and an explicit install/permission manifest, and you verify there are no unexpected network calls or credential exfiltration, the inconsistencies will be less concerning. Until then, treat the skill with caution.

Type: OpenClaw Skill Name: sql-database-toolkit Version: 1.0.0 The SQL Database Toolkit is a comprehensive data analysis suite, but it contains several high-risk security vulnerabilities. Most notably, 'file_connector.py' includes a loader for Pickle files using 'pd.read_pickle', which is a well-documented vector for Remote Code Execution (RCE) if an agent is tricked into loading a malicious file. Furthermore, the 'query' method in 'file_connector.py' implements a custom SQL parameter replacement logic by manually swapping strings (e.g., replacing ':key' with 'value'), which is highly susceptible to SQL injection attacks compared to standard parameterized queries. While these appear to be unintentional architectural flaws rather than overt malice, they represent a significant attack surface for prompt-injection-based exploits against the AI agent.