The Agent Skills Directory

[SAFE]: The skill is designed for complex financial tasks and adheres to industry-standard modeling practices. It utilizes legitimate Python libraries (pandas, openpyxl) and a local validation utility (./scripts/Xlsx) to automate data analysis and reporting without exhibiting any malicious behaviors.
[PROMPT_INJECTION]: The skill includes an inherent surface for indirect prompt injection (Category 8) as it is designed to ingest and process data from external sources such as web searches, APIs, and user-provided spreadsheets. However, the risk is significantly mitigated by the skill's mandatory validation protocol, which requires the agent to run structural and formula checks (recheck, reference-check, validate) before finalizing any output.
Ingestion points: Data from web_search, datasource, and user-uploaded Excel/CSV attachments (identified in SKILL.md).
Boundary markers: Absent; the instructions do not specify explicit delimiters for interpolated data.
Capability inventory: Execution of the local CLI tool ./scripts/Xlsx via shell and Python code execution via ipython (identified in SKILL.md).
Sanitization: The skill relies on structured validation routines to detect and reject malformed formulas and structure errors before delivery.
[DATA_EXFILTRATION]: The skill enforces a mandatory source citation rule for all external data. This requires the inclusion of 'Source Name' and 'Source URL' columns in deliverables, providing transparency and auditability for all externally sourced information.

xlsx