serper-search

**Quick judgment:** A solid, focused web/news search skill that reliably exposes real Google results via Serper.dev. Well-suited as a general-purpose search primitive for agents that need fresh, source-linked information, not synthesized summaries. **Strengths** - Real Google SERP + news data with links, snippets, knowledge graph, and related questions in one tool call. - Clean, minimal interface (`query`, `num`, `searchType`) that agents can use without complex prompting. - Good documentation and clear setup instructions (API key via env or config, default result count). **Risks / Limitations** - **External dependency:** Entirely dependent on the Serper.dev API and network; outages, latency, or rate limits can break or degrade the skill. - **Content risk:** Returns arbitrary web/news content (including possibly harmful, biased, or NSFW material); calling agents must apply their own safety and filtering. - **Scope:** Only search/news retrieval; no built-in ranking customization, language filters, or post-processing beyond what Serper provides. **Recommended scenarios** - Use as the default web/news search tool when you need up-to-date information, links, and snippets as grounding. - Combine with summarization/analysis tools to turn raw SERP output into user-friendly answers. - Prefer in research, fact-checking, and citation-heavy workflows where direct sources are required rather than purely model-generated responses.