The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and normalize untrusted data from various formats (Excel, CSV, PDF, Word) which serves as a potential vector for indirect prompt injection.
Ingestion points: Data enters the agent's context through the read_source.py script output and direct file reads of user-supplied documents.
Capability inventory: The skill utilizes file system access (read/write) and execution of local Python scripts.
Boundary markers: The instructions explicitly require the agent to summarize mappings and counts for user confirmation before proceeding to the next pipeline step, which acts as a manual verification boundary.
Sanitization: The script converts input to strings and joins them with delimiters, but does not sanitize content against embedded natural language instructions.
[COMMAND_EXECUTION]: The skill instructions direct the agent to run a bundled Python script read_source.py using the shell. This is a legitimate functional requirement for processing binary Excel files and ensuring UTF-8 encoding consistency across different operating systems.
[EXTERNAL_DOWNLOADS]: The skill documentation notes a dependency on openpyxl, a standard and well-known Python library for spreadsheet manipulation. The instruction for users to pip install this package is a standard development practice for skills requiring specific file format support.

github-extract-findings