The Agent Skills Directory

[SAFE]: Secret Management: The skill correctly utilizes environment variables via a require_env helper to handle the BIGDATA_API_KEY, ensuring that sensitive credentials are not hardcoded or leaked. It explicitly implements a 'no fallback' policy for secrets.
[SAFE]: Dependency Integrity: The skill relies on the official bigdata-client Python SDK for its core functionality. It provides clear instructions for installation in an isolated environment, which is a standard and safe development practice.
[SAFE]: Network Security: Outbound network traffic is directed exclusively to the official api.bigdata.com domain and authenticated AWS S3 presigned URLs for batch search features. These are legitimate and well-known endpoints for the described service.
[SAFE]: Code Transparency: The implementation is clear, modular, and extensively documented (including dual-language documentation). No evidence of obfuscation, hidden URLs, or unauthorized data exfiltration was found during the analysis.
[SAFE]: Indirect Prompt Injection: While the skill processes external news and financial data, which is an inherent risk factor for indirect injection, this is its primary intended function. The toolkit provides raw data for agent processing as requested, and the behavior is consistent with its stated purpose.

bigdata-skill