Expert Web Research

Most AI agents research like interns. They grab the first 3 Google results, summarize them, and call it done. This skill turns your agent into a senior analyst.

Built from hundreds of real research tasks across product validation, market analysis, health/clinical evidence, travel planning, and technical buying decisions. Every principle here comes from making actual high-stakes decisions, not theory.

What changes when your agent has this skill:

• Defines a research contract before searching (objective, scope, required outputs)
• Uses a 5-tier source hierarchy: primary/official > peer-reviewed > independent benchmarks > first-hand user evidence > marketing claims
• Runs 8-12 targeted queries across diverse evidence tiers instead of 2-3 generic searches
• Applies anti-noise controls: de-duplicates by factual equivalence, detects affiliate/agenda bias, catches cherry-picking
• Domain-specific rules for health/medicine, travel, product/SaaS, and engineering decisions
• Evidence extraction format with explicit confidence levels and contradiction mapping
• Source quality rubric scoring recency, methodology, independence, reproducibility, sample quality, bias risk
• Output truth controls: never treats marketing as fact, never invents data, states uncertainty clearly
• Closes every research task with recommendation tiers and next validation moves

Your agent will produce research that reads like it came from a paid analyst, not ChatGPT.

Works with any AI assistant that has web search and web fetch capabilities.