Agent
Thorough structured researcher. Can follow links and extract full page content (WebFetch). Strong multi-step reasoning, connects concepts across sources. Citation-aware. Best for: comprehensive research, following leads across pages, synthesis, technical depth.
Agent
Aggressive parallel searcher with Google grounding search. High quality, current results. Fast, handles large result sets. Best for: broad coverage, current events, parallel multi-angle search, rapid fact-gathering.
Agent
Quick factual lookups and common knowledge. Good for wide-range info gathering efficiently. NOT suited for deep analysis. Best for: current factual information, common knowledge verification, efficient wide-range info gathering.
Agent
Real original content extraction focus. Multi-source original info extraction via extract tool after finding relevant URLs. Best for: multi-source validation, consensus finding, getting complete original content rather than summaries.
MCP server Claude CodeCodexCursor
PRISM: An alternative to MCMC for rapid analysis of models. Runs locally from the prism Python package.