The Agent Skills Directory

[COMMAND_EXECUTION]: The dispatcher script scripts/search.py uses subprocess.run to call provider-specific search scripts. It passes arguments as a list rather than a string, which is a secure practice that prevents shell injection attacks.
[CREDENTIALS_UNSAFE]: The skill correctly avoids hardcoding sensitive information. API keys for Google and Bing are retrieved from environment variables, ensuring that credentials are not exposed in the source code.
[EXTERNAL_DOWNLOADS]: The skill performs legitimate network requests to established search provider APIs (Microsoft, Google, and DuckDuckGo). These operations are required for the skill's primary functionality and target trusted domains.
[PROMPT_INJECTION]: Like all search tools, this skill ingests untrusted data from the internet. While it does not implement specific boundary markers for search results, the risk is typical for this category of tool and is not escalated by any dangerous code execution or data exfiltration logic.

web-search