research-lookup

This skill acts as an intelligent research orchestrator, designed for scientists, analysts, and students who require precise, high-quality information retrieval. By implementing an automated backend routing system, the skill evaluates the nature of your input to select the most appropriate engine: parallel-cli search for high-speed general web data and academic prioritization, the Parallel Chat API for comprehensive, multi-source deep research synthesis, or Perplexity sonar-pro-search when precise scholarly paper verification is required.

• Intelligent Routing: Automatically detects academic intent, deep synthesis requirements, or general data needs to optimize for both latency and depth.

• Academic Source Prioritization: Specifically configured to query high-impact repositories such as PubMed, arXiv, Semantic Scholar, and Nature to ensure evidence-based results.

• Multi-Backend Integration: Seamlessly manages parallel-cli, Parallel Chat API, and OpenRouter-based sonar-pro-search within a unified workflow.

• Visual Augmentation: Integrated with the scientific-schematics skill to generate publication-quality diagrams and figures from research text, enhancing the overall quality of your output documents.

• Flexible Overrides: Allows users to manually force specific backends (parallel, perplexity, or fast-search) for specialized workflows or benchmarking.

• Ideal for literature reviews, fact-checking technical specifications, gathering market intelligence, or searching for DOI-indexed articles and peer-reviewed studies.

• Best practices include conducting dual-stage searches (academic vs. general) for critical scientific inquiries to ensure both scholarly depth and real-time news coverage.

• Inputs typically include specific research questions, citation requests, or methodology inquiries, while outputs provide synthesized reports, annotated findings, or structured JSON data files.

• Practical constraints: Users should manage API keys for specific backends (PARALLEL_API_KEY, OPENROUTER_API_KEY) and be aware of varying latency profiles, where deep research may take up to 5 minutes to complete while parallel-cli offers near-instant results.