The Agent Skills Directory

[SAFE]: Detailed analysis of the skill reveals no evidence of malicious behavior, obfuscation, or security vulnerabilities. The implementation relies exclusively on the Python standard library, minimizing the attack surface.
[EXTERNAL_DOWNLOADS]: The skill facilitates network connections to 12 well-known search providers, including Google, Anthropic, OpenAI, Brave, and Tavily. These connections target official API endpoints for the intended purpose of web searching.
[PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from the web (search snippets and titles).
Ingestion points: Data is fetched from remote search APIs via urllib.request in scripts/web-search.
Boundary markers: Results are returned in a structured JSON format (combined.json) and printed with distinct CLI headers to separate results from different providers.
Capability inventory: The skill possesses the capability to write to the system's temporary directory and perform network requests to configured APIs.
Sanitization: The skill normalizes various API response formats into plain text and utilizes flags like no_html=1 for providers such as DuckDuckGo to reduce the presence of executable tags.

web-search