The Agent Skills Directory

[SAFE]: The skill implementation uses standard Node.js libraries for web scraping and content parsing. No signs of malicious behavior, obfuscation, or unauthorized data access were found.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by fetching untrusted external data from search results and specific URLs and returning it to the agent. Ingestion points: Network fetch operations in content.js and search.js. Boundary markers: None present to distinguish search content from instructions. Capability inventory: Limited to network requests and terminal output; no command execution or persistent file storage. Sanitization: Employs @mozilla/readability and JSDOM to extract article content and Turndown for markdown transformation.
[EXTERNAL_DOWNLOADS]: The skill performs legitimate network requests to search.brave.com and other external domains to fulfill its search and extraction tasks. This behavior is consistent with the primary use case of a web search tool.

brave-search