What problem does it solve? When researching technical topics, answers often come from outdated blog posts or unverified secondhand claims. This Skill enforces a disciplined investigation process that traces every claim back to primary sources—official documentation, source code, specifications, and first-party APIs—so decisions rest on verifiable evidence. ## Core Features & Use Cases - Parallel subagent delegation: When subagent tooling is available, splits the question into independent sub-questions and dispatches 4–10 parallel researchers, each returning evidence with verbatim quotes and URLs rather than conclusions. - Direct investigation fallback: When subagents are unavailable, performs the research itself by searching and reading primary sources directly, without spawning background processes. - Source discipline: Prioritizes current first-party sources (treating AI/dev information older than a year as potentially outdated), checks local facts first (code, git history, installed dependencies), and marks anything unverifiable as "unconfirmed." - Use Case: You need to decide which API version a library supports. The Skill dispatches researchers to the official docs, changelog, and source repository, then returns cited conclusions—or explicitly flags what could not be confirmed. ## Quick Start Ask the agent to research a technical question using the research skill, for example: "Research how the latest version of this framework handles streaming responses and cite primary sources."