The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external S&P Global APIs to generate takeaways. This data could potentially contain malicious instructions if the external source were compromised. * (1) Ingestion points: get_rounds_of_funding_from_identifiers and get_company_summary_from_identifiers tools fetch external content. * (2) Boundary markers: No specific delimiters or safety instructions wrap the ingested text. * (3) Capability inventory: The skill can install npm packages, write files, and present files. * (4) Sanitization: No explicit sanitization of the fetched data is performed.
[EXTERNAL_DOWNLOADS]: The skill automates the installation of third-party dependencies from the NPM registry. Evidence: Step 6 commands the agent to execute npm install simple-icons sharp.
[COMMAND_EXECUTION]: The skill executes multiple shell commands for environment setup and quality assurance. Evidence: Execution of npm install, python -m markitdown, and soffice.py for conversion.
[DYNAMIC_EXECUTION]: The skill uses dynamic script generation to process brand icons and build the final PowerPoint file. Evidence: The skill provides JavaScript logic using the sharp and pptxgenjs libraries for the agent to execute.

funding-digest