Category ranking

Best Research Agent Skills in 2026

Research skills impose structure on open-ended investigation: decompose the question, gather sources, cross-check claims, and produce a cited summary. Without one, agents tend to stop at the first plausible answer.

Pair these with a web-search or fetch MCP server for best results — the skill provides the method, the server provides access to the material.

239个结果
ES
emilkowalski/skills
研究分析

设计师和工程师的技能。

5.4万75
JM
jakubkrehel/make-interfaces-feel-better
研究分析

一个能让你的界面更舒服的特工技能。

391859
JB
jinchenma94/bazi-skill
研究分析

四柱八字命理分析

377358
HT
hotcoffeeshake/tong-jincheng-skill
研究分析

童锦程视角 Skill — 用深情祖师爷的思维框架分析人际关系

297672
LD
larashero3-dotcom/lieflat-gongwen
研究分析

通过102万字语料提炼的公文写作 skill,把公文写作风格变成可测量、可验收的量化数据

228258
EH
everettfish/holo-card-studio
研究分析

Turn the user's description or uploaded reference into a finished, editable Blender card and an interactive Three.js page. Preserve the requested subject, style, typography and destination.

204654
CB
crazyykhllc-bit/cyberppt
研究分析

一个用于生成高密度、可编辑、咨询风格 PowerPoint 的 Codex Skill,支持 SCR 叙事、风格确认和 PPTX 质量检查。

185858
RG
robmitt/grill-me-skill
研究分析

Robmitt/Grill-Me-Skill 是从 https://github.com/robmitt/grill-me-skill. 导入的特工技能候选人。

138729
DC
dimillian/codexskillmanager
研究分析

macOS 应用来管理你的 Codex 技能

137845
WC
wangjunqing-coder/huasheng13-skill
研究分析

wangjunqing-coder/huasheng13-skill 是从 https://github.com/wangjunqing-coder/huasheng13-skill. 导入的 Agent Skill 候选人。

109347
MF
mrtooher/fable-mode
研究分析

mrtooher/fable-mode是从https://github.com/mrtooher/fable-mode.导入的特工技能候选人。

88455
XC
xbuilderlab/cheat-on-money
研究分析

xbuilderlab/cheat-on-money是从[[T0]导入的特工技能候选人。

77645
查看全部匹配条目 · 239

FAQ

Do research skills work without a search MCP server?

They can work from documents you provide, but they shine when the agent can fetch sources itself. Combine a research skill with a search or fetch server from the scenario pages below for a complete setup.

How do I trust the citations an agent produces?

Choose skills that require inline citations tied to fetched URLs, then spot-check. Skills that separate "claims" from "sources" in their output format make verification much faster.

Research skill versus deep-research products?

Dedicated deep-research products are heavier and slower; a research skill runs inside your existing client, costs only your normal model usage, and you can edit the workflow to your needs.

相关指南