The Agent Skills Directory

[SAFE]: The skill employs standard data processing tools and libraries for its intended purpose. All command executions and scripts are local to the skill's environment or target well-known Python packages. The workflow includes mandatory validation steps (recheck, validate) to ensure the integrity of the generated files.
[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it is designed to ingest data from external sources such as web searches and APIs. However, the risk is inherent to the data analysis use case and is mitigated by structured output requirements and mandatory source citations. Evidence chain:
Ingestion points: External data fetched via web_search, datasource, or API calls used to populate spreadsheet cells (documented in SKILL.md).
Boundary markers: None explicitly defined for untrusted data within Excel cells, though source citation columns are mandatory.
Capability inventory: Python execution via ipython and shell command execution via the shell tool for the KimiXlsx CLI utility.
Sanitization: Not explicitly described for external content before interpolation into the spreadsheet.

kimi-xlsx