The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run in tools/mcp_query.py to execute the curl command. This is used to communicate with a remote MCP server. The implementation uses a list of arguments and json.dumps() for the request payload, which follows security best practices to prevent shell injection.
[EXTERNAL_DOWNLOADS]: The skill relies on the external akshare library for most of its data retrieval. It also fetches real-time financial data, news, and announcements from several external sources, including a hardcoded MCP server IP (82.156.17.205), Caixin news, CCTV news, and aastocks.com.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from external financial news and company announcements.
Ingestion points: Data is pulled from external websites in tools/stock_news.py, tools/market_news.py, and tools/stock_announcements.py.
Boundary markers: The skill does not implement specific delimiters or warnings for the agent to ignore instructions within the fetched data.
Capability inventory: The skill is limited to data retrieval and display. It does not have capabilities for file writing, persistent storage, or privileged command execution based on the ingested data.
Sanitization: tools/market_news.py truncates summaries and tools/stock_announcements.py uses keyword-based filtering to select relevant content.
[REMOTE_CODE_EXECUTION]: The skill interacts with a remote MCP server at http://82.156.17.205/cnstock/mcp. While it calls specific tools on this server (brief, medium, full) rather than executing arbitrary code, the skill's functionality is dependent on this external third-party infrastructure.

finance-data