The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/batch_regenerate_word.py and the main SKILL.md workflow use Python's subprocess module and shell commands to orchestrate execution between various internal scripts. While commands are generally constructed as lists (reducing shell injection risk), they execute logic based on user-controllable parameters like ASIN and category names.
[EXTERNAL_DOWNLOADS]: The skill documentation (SKILL.md and references/path-setup.md) identifies an optional dependency on a third-party tool named 'web-access' (from the repository github.com/eze-is/web-access). This tool is used for automated data extraction from the Sif website and is not maintained by a recognized trusted vendor.
[PROMPT_INJECTION]: The system ingests and processes data from user-provided Excel spreadsheets (.xlsx). This data is subsequently used as context for the AI agent to write descriptive analysis sections (insights) in Word and Markdown reports. This flow presents an indirect prompt injection surface where malicious data embedded in the spreadsheets could attempt to manipulate the agent's reporting behavior.

sif-keyword-scout