The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data sources (CSV, Excel, JSON) to generate reports. This establishes a surface where malicious instructions hidden within the input data could potentially attempt to influence the agent's output formatting or content.
Ingestion points: pandas.read_csv is used in the generate_report and create_sales_report functions in SKILL.md.
Boundary markers: The example code does not include specific delimiting or 'ignore instructions' markers for the input data.
Capability inventory: The skill utilizes file writing (with open(), plt.savefig()) and file reading capabilities to generate the reports.
Sanitization: No explicit data sanitization or validation logic is present in the boilerplate examples provided.
[COMMAND_EXECUTION]: The skill provides Python code snippets that interact with the local file system to read datasets and save generated visualizations and report files (HTML/PNG).

report-generator