xlsx
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEPERSISTENCEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PERSISTENCE]: The
scripts/recalc.pyscript writes a StarBasic macro to the LibreOffice user configuration directory (Standard/Module1.xba). This modification persists across system sessions and is automatically triggered whenever the agent uses the recalculation tool. Note that the script's installation method may overwrite existing user macros within the same module file. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and manipulate untrusted Excel and CSV files using
pandasandopenpyxl. - Ingestion points: Data is loaded into the agent context via
load_workbookandpd.read_excelas documented inSKILL.mdandscripts/recalc.py. - Boundary markers: The instructions lack explicit boundary markers or instructions for the agent to ignore potentially malicious instructions embedded within spreadsheet cells.
- Capability inventory: The skill grants the agent the capability to execute shell commands (via the
recalc.pyhelper) and perform file system writes to save spreadsheet outputs. - Sanitization: There is no evidence of sanitization or content validation performed on the data read from spreadsheets before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill instructions and internal scripts rely on the
subprocessmodule to invoke external binaries, including the Python interpreter and thesoffice(LibreOffice) executable. While these are used for the intended purpose of formula recalculation, they provide the agent with a mechanism to interact with the underlying operating system.
Audit Metadata