
from claude-ai-music-skills180
Validate research outputs: check source accessibility, verify quotes, cross-check dates and facts, archive sources, and produce a verification report for human
Research Verifier is a quality-control skill that validates research gathered by other agents. It checks URL accessibility and archival copies, verifies that quotations are verbatim, ensures date consistency across sources, cross-references factual claims, and produces a structured verification report highlighting critical issues and warnings.
Run this after research collection and before human editorial review or publication — when RESEARCH.md and SOURCES.md are ready. It's intended to be the last automated check before handing materials to a human verifier.
Designed for agents that can read files and perform web checks (agents with Read, WebFetch/WebSearch, and file access). Works well in Claude Code plugin contexts and other agent runtimes that support tool usage for web and file I/O.
Research Verifier is a prompt-only skill (no scripts) for fact-checking and citation validation in documentary music research projects. It provides a thorough verification checklist and report template but relies entirely on agent compliance rather than enforceable logic. The SKILL.md is well-structured with clear frontmatter, allowed-tools restrictions, and references to supporting files (checklists.md, patterns.md) that aren't bundled, making it dependent on its parent repo context.
Clean skill with no security concerns. Restricted tool access in frontmatter is a positive signal. Main limitations are: (1) no scripts to enforce verification steps, (2) supporting reference files not included in skill bundle, (3) narrow domain focus on music documentary research.