research-lookup

The research-lookup skill functions as an automated research assistant designed for scientists, analysts, and developers who require high-precision information retrieval. By implementing an intelligent backend routing mechanism, the skill directs your queries to the most appropriate engine: parallel-cli search for high-speed, cost-effective general and technical web queries; the Parallel Chat API for comprehensive, multi-step deep research; or Perplexity sonar-pro-search for targeted academic paper and literature discovery. This system is engineered to handle complex tasks such as background research for scientific writing, cross-referencing industry statistics, verifying technical specifications, and identifying seminal academic papers. It bridges the gap between quick informational lookups and exhaustive systematic literature reviews by optimizing the underlying search architecture based on user-defined constraints and input keywords.

• Intelligent Routing: Automatically selects between fast web search, deep synthesis, or specialized academic search based on content markers.

• Academic Prioritization: Enhances results for scholarly queries by targeting high-impact domains including PubMed, ArXiv, Nature, Science, and IEEE.

• Multi-Backend Versatility: Seamlessly switches between the parallel-web infrastructure, the Parallel Chat core model, and Perplexity sonar-pro-search via OpenRouter.

• Citation and Verification: Facilitates the identification of DOIs, peer-reviewed articles, and systematic reviews to support evidence-based decision-making.

• Scientific Visual Integration: Offers hooks to incorporate AI-generated schematics and publication-quality diagrams via scientific-schematics integration.

• Extensive Filtering: Supports specific metadata constraints like date ranges (after-date) and domain inclusion/exclusion lists.

• Inputs/Outputs: The skill processes natural language research queries and returns synthesized data, formatted citations, or raw JSON research outputs for downstream agent processing.

• Practical Tips: For best results, use academic keywords like 'meta-analysis', 'DOI', or 'systematic review' to trigger specialized academic routing. When conducting complex research, consider the 60s-5min latency for deep research modes.

• Constraints: The effectiveness depends on available API keys (PARALLEL_API_KEY, OPENROUTER_API_KEY). Manual overrides are available for users who need to force specific search backends.

• Workflow Integration: Designed for use within the K-Dense Agent Skills ecosystem, compatible with tools like Cursor, Claude Code, and other standard-compliant AI agents.