The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's HTML generation process includes references to external visualization and mathematical rendering libraries.
The reports include script tags for echarts and mathjax hosted on the cdn.jsdelivr.net Content Delivery Network. These are well-known, established services used for professional data visualization.
[COMMAND_EXECUTION]: The skill requires the agent to execute local Python scripts and shell commands to facilitate its primary function.
It runs stock_full_report.py to gather market data from financial portals via the akshare library.
It uses npx commands to initiate MCP (Model Context Protocol) servers for searching industry trends and news via Zhipu AI and Tavily APIs.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes external news and research data.
Ingestion points: News and research reports are read from output/data_<stock_code>.json and MCP search results during the Phase 2 analysis stage.
Boundary markers: Absent; there are no instructions for the agent to use specific delimiters or ignore instructions that might be embedded in the fetched news or reports.
Capability inventory: The agent can write reports to the local output/ directory, read local template files, and perform shell-based file concatenation.
Sanitization: No explicit sanitization or filtering of the ingested news titles or research text is performed before processing.

stock-analysis-github