What problem does it solve? Starting a coding or optimization task without domain knowledge leads to weak first attempts and repeated dead ends. This Skill structures the pre-coding research phase: it guides broad-to-focused web searches, preserves raw sources for verification, and produces synthesized research notes with an index so findings persist across sessions. ## Core Features & Use Cases - Structured research workflow: Survey 3+ approaches, compare 2-4 candidates on evidence and complexity, then commit to one before writing code. - Immutable raw source archive: Capture web articles, papers, GitHub repos, videos, threads, docs, and chat logs into notes/raw/ with per-type frontmatter conventions. - Failure-mode recovery: Diagnose and recover from blocked WebFetch calls, empty search results, contradictory sources, stale notes, and underexplored topics. - Grounding verification: Optionally spawn a Synthesis Reviewer subagent that verdicts every claim in a note as grounded, partially-grounded, inferred, contradicted, or unverifiable against its cited sources. - Use Case: Before implementing a CVRP solver, use this Skill to survey state-of-the-art heuristics, save the key papers to notes/raw/, write a comparison note in notes/research/, and pick an approach backed by benchmark evidence. ## Quick Start Ask the agent to research the problem domain before coding, saving raw sources and writing structured research notes with an updated index.
