The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest data from external sources, such as news articles and earnings transcripts, and render this content into an HTML report. If the retrieved data contains malicious formatting or code, it could influence the report's presentation or the agent's processing.
Ingestion points: Data enters the skill context through Kensho Grounding and S&P Global MCP tool results.
Boundary markers: The instructions do not explicitly require the use of delimiters or specific warnings for the agent to ignore instructions embedded within the ingested data.
Capability inventory: The skill uses bash for managing intermediate files and opening the final HTML output.
Sanitization: There are no specific instructions for escaping or sanitizing the data retrieved from external searches before it is written into the HTML report template.
Command Execution Consideration: The skill parses a company ticker from user arguments and incorporates it into a filename for shell operations, such as creating the report and opening it. Using user-supplied input in shell parameters is a common area for review to ensure that only intended file operations are performed.
Dynamic Script Generation: The skill generates an HTML document that includes embedded JavaScript code and data gathered during the research phases. This file is subsequently opened in a browser environment to display interactive charts.
External Resource References: The HTML template loads the Chart.js library and its annotation plugin from a public Content Delivery Network (CDN). Referencing well-known CDNs for standard visualization libraries is a common practice for generating interactive reports.

earnings-preview-single