ShinyGua
18 verified skills
ar-debate
Generic debate loop — runs structured multi-role debate for high-stakes research decisions
testing
ar-method
Method agent — generates method cards, assesses method-question fit, identifies validity threats
data-ai
ar-paper-reviewer
Academic paper reviewer — scores paper on 6 dimensions, identifies blockers, generates revision briefs
research
ar-research-writer
Academic paper writer — researches topic via web and writes a full paper draft with track-specific structure
development
ar-retrieval
Corpus retrieval — searches academic, policy, media, and report sources to build a curated corpus
development
ar-citation-verifier
Citation verifier — checks that in-text citations accurately represent what the cited source actually says
testing
ar-claim-builder
Claim builder — constructs claims from evidence, builds argument tree, detects overclaims, triggers debate for rival explanations
development
ar-scoping
Topic scoping — refines topic seed into research question, scope boundaries, and feasibility assessment
research
ar-progress-monitor
Background progress watcher — reads status.json and updates task list checklist in real-time
testing
ar-evidence
Evidence agent — extracts evidence units from reading notes, groups by theme, identifies contradictions
data-ai
ar-init
Interactive project initialization — asks user for research topic and track, creates workspace, launches pipeline
devops
ar-lit-synthesis
Literature synthesis — clusters sources into themes, maps debates, identifies gaps, and compares schools of thought
tools
ar-orchestrator
Research pipeline orchestrator — manages all 9 stages with 5 human approval gates
devops
ar-pdf-exporter
Academic document exporter — converts markdown paper draft to PDF and Word (.docx)
documentation
ar-reader
Reader agent — reads individual sources, extracts structured notes, quotes with spans, and positions in debate
data-ai
ar-ref-checker
Reference checker — verifies that all references in the paper exist, with correct author names, titles, and years
testing
ar-theory
Theory agent — proposes theoretical frameworks, ranks by fit, and identifies rival theories
development
ar-source-auditor
Source auditor — classifies, scores reliability, verifies metadata, and recommends inclusion/exclusion for corpus sources
testing