Perplexity Search Skill

**Quick judgment**: A solid, focused web search skill that cleanly exposes Perplexity’s Search API with sensible defaults and recency filters. It’s well-suited as the “live web search” component in OpenClaw workflows, especially where current information is essential. **Strengths & behavior** - Real-time, ranked web results with optional recency constraints (day/week/month/year). - Configurable result count (1–10) and option to return raw JSON for programmatic use. - Clear CLI-style interface and straightforward API key configuration via environment variables/OpenClaw config. - Response structure (title, url, snippet, date, last_updated) is practical for downstream processing. **Risks & limitations** - **External dependency**: Completely dependent on Perplexity’s API availability, rate limits, and quality of search results. - **API key & cost**: Requires a Perplexity API key; usage is billable ($5 / 1,000 requests), so high-volume or iterative queries may incur non-trivial cost. - **Rate limiting**: Workflows that fan out many searches may hit Perplexity rate limits; error handling exists but cannot remove this constraint. - **Unvetted content**: Returns arbitrary web pages; any downstream automation that “trusts” snippets or links must treat them as untrusted input. - **Feature scope**: Only basic filters (count, recency); no advanced search operators or rich filtering (e.g., site/domain filters) exposed here. **Recommended scenarios** - Use when you explicitly need **fresh or time-sensitive information** (news, regulatory updates, fast-changing markets). - Use as a **fallback search provider** when other search skills (e.g., Brave search) are unavailable or underperforming. - Use in **research and analysis pipelines** that collect URLs/snippets for later scraping, summarization, or clustering. - Avoid or tightly budget usage in **high-volume loops** where cost and rate limits could become problematic, and avoid using results as trusted input without validation.