forge-research

What problem does it solve?

forge-research prevents confident-but-shaky outputs by forcing AI or humans to perform real research: decomposing questions, prioritizing source quality, cross-checking claims, and explicitly stating confidence and unknowns.

Core Features & Use Cases

• Question decomposition: turns a broad topic into searchable sub-questions so each step reduces uncertainty.
• Source hierarchy with recency: prioritizes primary sources over secondary and tertiary, and favors newer material in fast-moving domains.
• Search-vs-ask decisioning: searches for public answers, asks for private/contextual answers, and reads code when the question is codebase-specific.
• Anti-hallucination verification: requires cited links and verified quotes, plus counter-searches for consequential conclusions.
• Confidence-based synthesis template: outputs claims labeled as established, likely, disputed, and includes open questions plus a method section.

Quick Start

Use forge-research to answer this question by decomposing it into sub-questions, performing searches with a primary>secondary>tertiary source hierarchy, verifying cited URLs and quotes, and returning the full template with confidence levels and open questions.