科研 Skill

材料科学文献检索

materials-lit-search

Covers formulating literature search queries for materials science questions and synthesizing/citing results responsibly. Requires a web search tool; it cannot substitute for one.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Materials Science Literature Search

Overview

This skill is knowledge about how to search and synthesize materials-science literature well — it is not itself a search capability. An agent following this skill still needs an actual tool (web search, a database API call) to execute against; without one, the best this skill can offer is documentation, not results.

When to use this skill

Activate when the request involves:

  • Literature search, prior work, related studies, has this material/property been reported before
  • "What does the literature say about X," "find papers on...," "has this synthesis route been tried"

Core usage

Query formulation

Weak query:  "good battery cathode material"
Better:      "high capacity cathode material lithium ion battery cycling stability"
Best (using materials-specific terminology and property qualifiers):
             ("cathode material" OR "positive electrode") AND "Li-ion battery"
             AND ("capacity retention" OR "cycling stability") AND ("layered oxide" OR "NMC")

Specific composition notation, phase names, and property terminology (not lay paraphrases) retrieve substantially more relevant results — materials science literature searches benefit especially from including both a material's common name and its formula/composition notation, since papers vary in which they emphasize.

A systematic search checklist, not a single query

  • Terminology: both common names and formal composition/structure notation (e.g. "NMC811" and "LiNi0.8Mn0.1Co0.1O2")
  • Study type: distinguish computational (DFT-predicted) from experimental (synthesized and measured) results explicitly — a materials-science literature search that doesn't separate these can conflate a predicted-but-unsynthesized candidate with a validated result
  • Recency: materials databases and reported "best" properties for a given application (e.g. battery capacity records) change quickly; check whether a cited value is still representative of the current state of the art

Synthesis discipline

  • State what search was actually performed (terms, database, date range) so a reader can assess coverage.
  • Distinguish a computational prediction from an experimentally validated result — this distinction matters more in materials science than in many fields, since a large fraction of the high-throughput screening literature reports predicted-but-unsynthesized candidates.
  • Note contradictory findings (e.g. conflicting reported values for the same material's property, common when measurement conditions differ) rather than citing only the most favorable value found.

Validation & Pitfalls

  • A literature search is only as good as the search tool's actual coverage and the agent's actual ability to invoke it — if the executing agent has no real search/API access, its output is recalled/memorized training data dressed as a search result, not a current search, and should never be presented as if a search was actually performed when it wasn't.
  • Citation fabrication is a known failure mode for language models operating without real retrieval — a plausible-sounding author/year/journal citation is not evidence it's real; any citation this skill (or any skill in this repo) produces should be verifiable against an actual source.
  • Reported material properties in the literature vary with measurement/synthesis conditions that aren't always stated prominently (sample purity, measurement temperature, processing history) — a single literature-reported value should be treated as one data point under specific conditions, not a universal material constant, especially for structure-sensitive properties (mechanical strength, ionic conductivity) that vary substantially with microstructure.
  • Predicted (computational screening) results significantly outnumber experimentally validated ones in current materials discovery literature — a search that doesn't distinguish these risks presenting an unsynthesized computational prediction with the same confidence as a validated experimental result.

示例提示词

帮我构建高效的检索式查找关于钙钛矿稳定性的最新文献并整理引用。

所属插件

材料表征、性能与材料信息学面向实验与数据驱动的材料研究:XRD/电镜/谱学/热分析/力学测试等表征,合成与加工,电化学与半导体表征,相场、CALPHAD 与连续介质力学,材料数据库、特征工程、贝叶斯优化与主动学习,选材与生命周期评价。

安装该插件会一并获得同领域的 42 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

材料科学文献检索能做什么?

Covers formulating literature search queries for materials science questions and synthesizing/citing results responsibly. Requires a web search tool; it cannot substitute for one。

如何安装材料科学文献检索?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「材料科学文献检索」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

材料科学文献检索需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

材料科学文献检索经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。