| name | literature-gap-map |
| description | Build an evidence-bounded map of unresolved research questions, boundary conditions, and missing comparisons from existing ResearchOS literature artifacts. Use before ideation or a related-work section when a researcher needs to distinguish retrieval gaps from evidence-supported research opportunities. |
| tools | ["read_file","list_files","grep_search","write_file","finish_task"] |
| strict_tools | true |
| model_tier | standard |
| temperature | 0.15 |
| allowed_read_prefixes | ["user_inputs/literature-gap-map/","literature/","ideation/","_runtime/skill_sessions/"] |
| allowed_write_prefixes | ["literature/"] |
| outputs_expected | {"gap_report":"literature/skill_gap_map.md","gap_record":"literature/skill_gap_map.json"} |
| interaction | {"mode":"guided","language":"zh-CN","summary":"从已有笔记、综合和比较表中识别被证据支持的未解问题、边界条件和缺失对照，并区分“尚未检到”与“可主张研究缺口”。","request_required":true,"request_prompt":"请说明要定位的主题、目标决策（综述、Idea、实验或 Related Work）和不应被误当成研究缺口的范围。","example_request":"基于这些笔记定位 treatment heterogeneity 的机制边界和缺失 baseline；将检索覆盖不足与真正可进入 T4 的问题分开。","required_inputs":[{"id":"evidence_sources","label":"文献证据来源清单","description":"一行一个 workspace 相对路径，指向 paper notes、PDF note cards、synthesis、比较矩阵或已核验记录；建议至少 3 项。","paths":"[Truncated]","extensions":"[Truncated]","min_bytes":20,"example":"user_inputs/literature-gap-map/sources.md"}],"optional_inputs":[{"id":"gap_scope","label":"缺口定义与使用场景","description":"可选；给出研究问题、目标场景、理论/方法/实验边界以及未来会如何使用该 map。","paths":"[Truncated]","extensions":"[Truncated]","min_bytes":10,"example":"user_inputs/literature-gap-map/context.md"}],"outputs":[{"id":"gap_report","label":"证据约束研究缺口图","path":"literature/skill_gap_map.md","description":"将可观察的不足、证据强度、反例/边界、需补检 section 和可进入 Idea 的问题分别列出。"},{"id":"gap_record","label":"缺口图结构记录","path":"literature/skill_gap_map.json","description":"机器可读的 gap 候选、来源/section 锚点、分类、证据等级、反证风险和推荐动作。"}]} |
Literature Gap Map
First validate that every listed source exists. Read the supplied materials by decision relevance and retain a path plus section or field anchor for every gap candidate. Use
grep_search only to locate supporting passages; read the relevant source before recording a conclusion. Do not interpret a missing note, a metadata-only record, or limited retrieval coverage as proof that a scholarly gap exists.
For each entry, classify it as an observed limitation, an untested boundary condition, a missing comparison, an unresolved mechanism, a retrieval/coverage deficiency, or unsupported speculation. Record supporting and countervailing evidence separately, the strongest wording currently allowed, what exact additional paper/section/evidence would upgrade it, and whether it is suitable for background, a T4 candidate input, or only further search.
Write both declared outputs. The report may suggest targeted follow-up queries or paper-note sections, but must not turn a gap map into a novelty claim or final hypothesis. Missing or contradictory evidence is a visible result, not a reason to invent a research story.
