The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external Excel files which could contain malicious instructions.
Ingestion points: Multiple commands in scripts/excel.py (specifically cmd_read, cmd_cell, and cmd_find) retrieve and display data directly from cell values in user-provided .xlsx workbooks.
Boundary markers: The instructions do not define clear delimiters or provide the agent with guidance to ignore potential instructions embedded within the spreadsheet data.
Capability inventory: The skill has broad capabilities to interact with the file system, including creating, writing, and editing files via cmd_create, cmd_write, and cmd_edit across the entire workspace.
Sanitization: The Python script does not perform any sanitization, filtering, or validation of cell content before it is passed to the agent's context.

excel