DuckDuckGo Web Search (1)

**Verdict:** A solid, no-API-key DuckDuckGo search skill that is well-suited as a backup or complementary web search tool in OpenClaw workflows, especially when primary search tools are unavailable. **Strengths:** - Uses DuckDuckGo via the `ddgs` library, avoiding the need for any API keys or paid services. - Simple interface: required `query` plus optional `count` (default 5, max 20), returning clean `title`/`URL`/`snippet` triplets. - Integrates naturally with `web_fetch` or similar tools for detailed page retrieval after search. **Key Risks / Limitations:** - Relies on unofficial scraping of DuckDuckGo; functionality may break if DuckDuckGo changes its frontend or if `ddgs` becomes outdated. - Subject to network instability and potential rate limiting under heavy use, which can degrade reliability. - Search results are biased toward English by default and offer limited control over region/language/customization. - As with any web search, responses can surface unsafe, biased, or low-quality content; no additional filtering beyond DuckDuckGo’s own behavior is evident. **Recommended Scenarios:** - Use as a **fallback or secondary search provider** when the primary `web_search` tool is missing, misconfigured, or rate-limited. - Use for **lightweight, small-batch web queries** where a handful of relevant URLs and snippets is sufficient. - Combine with `web_fetch` for **stepwise research workflows**, where the model first discovers candidate pages via this skill, then fetches and analyzes selected URLs. - Avoid relying on it as the **sole mission-critical search layer** in high-reliability environments due to inherent scraping and rate-limit risks.