Audit recent high-relevance papers from a target venue to identify claim-relevant experimental evidence gaps after core P0 results exist. Use for venue completeness and reviewer-risk analysis; not for redefining the research question or running experiments.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Run only when the Orchestrator authorizes the audit and P0 is substantially complete. Read research/research_state.yaml, the current Claim-Evidence status, docs/PROTOCOLS.md, docs/ROLE_HANDOFFS.md, and docs/EVIDENCE_COMPLETION_PROTOCOL.md.
Analyze 5-10 recent, highly relevant papers, prioritizing same task, research question, method class, venue and recency. Record search date, sources, relevance, inclusion/exclusion reasons and uncertainty. Stop once core expectations, material gaps and high-ROI actions are stable; do not expand into a systematic review. Compare datasets, baselines, metrics, seed/statistical practice, ablations, sensitivity, robustness, generalization, efficiency, complexity, qualitative analysis and figure/table conventions.
Produce an Evidence Gap Matrix with current status, venue frequency, Claim relevance, reviewer risk, estimated cost, information gain and recommended priority. Recommendations do not enter the queue automatically. A common but Claim-irrelevant experiment is P3/SKIP. Do not alter the Research Spine, Primary Claims or Method, and report any positioning concern only as risk.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Direct a claim-driven research experiment project from post-Idea intake through method realization, prioritized evidence collection, Human Gates, and Evidence Freeze. Use for scientific planning, state transitions, queues, budgets, or recovery; not for running code or independently auditing results.
日本語の概要は準備中です。原文の説明を表示しています。
Implement and execute approved research Experiment, Tuning, or Design Cards with reproducible logs, metrics, environment, and artifacts. Use for bounded coding, debugging, and runs; not for changing claims, priorities, budgets, or the Research Spine.
日本語の概要は準備中です。原文の説明を表示しています。
Validate and scientifically interpret completed research runs, including negative, null, boundary, confounded, failed, and inconclusive outcomes. Use after execution to assess evidence and claim impact; not to edit methods or schedule experiments.
日本語の概要は準備中です。原文の説明を表示しています。