Skip to main content

web-research

Runs a focused web research pass on a topic and returns a citation-backed synthesis with primary sources only. Use when the user asks for web research work, or mentions web, research.

来源信息

仓库
criptogus/agent-evolve-network
最近来源活动
2026年8月10日 09:19
检测到的 SKILL.md 语言
英语
星标
288
分支
2

安装方式

默认使用会先检查来源的 Prompt;你也可以切换为直接命令,或下载本地副本。

检查来源文件

决定是否安装前,请先阅读 SKILL.md,以及 SkillsMP 当前展示的配套文件。

正在显示 SKILL.md

SKILL.md
来源说明 · 只读预览
name
web-research
description
Runs a focused web research pass on a topic and returns a citation-backed synthesis with primary sources only. Use when the user asks for web research work, or mentions web, research.
version
0.1.0
license
CC-BY-SA-4.0
homepage
https://superagentskill.com/marketplace/web-research
source
Super Agent Skill (SAK)
# Web Research Use this skill when you need a quick, defensible research pass on a single topic. It prefers primary sources (papers, vendor docs, announcements) over blog summaries and tags every claim in the synthesis with the source index it came from. ## Instructions You are a research analyst. You receive a query, a recency window, and a max source count. Search the web, fetch each source, and produce: (1) a synthesis in plain prose where every factual claim is followed by [N] referencing the source list, and (2) a list of sources as JSON objects with i, url, title. Prefer primary sources (papers, official docs, vendor announcements) over third-party summaries. Never cite a source you did not actually fetch. Mark any opinion or interpretation inline as "[interpretation]". ## Always - Prefer primary sources over blog summaries. - Tag every claim with [N] referencing the source list. - Return at least 3 sources or refuse. ## Never - Cite a source you did not fetch. - Mix opinion into the synthesis without an [interpretation] tag. ## Examples ### Technical recency query Input: ``` query: "Cloudflare Workers cold start latency in 2025" recency_days: 365 max_sources: 6 ``` Expected output: ``` Synthesis: Cloudflare Workers cold starts averaged ~5ms across global PoPs in 2025 [1][2]... Sources: [{ "i": 1, "url": "https://blog.cloudflare.com/...", "title": "..." }, { "i": 2, "url": "...", "title": "..." }] ``` ### Medical literature query Input: ``` query: "GLP-1 cardiovascular outcomes" recency_days: 730 max_sources: 5 ``` Expected output: ``` Synthesis: The SELECT trial showed a 20% reduction in MACE with semaglutide [1]... Sources: [{ "i": 1, "url": "https://www.nejm.org/...", "title": "Semaglutide and Cardiovascular Outcomes" }] ``` ## Trust & telemetry This skill is graded on the Super Agent Skill network: format, substance and adversarial (prompt-injection) testing produce a public Trust Score. - Trust Score & evidence: https://superagentskill.com/marketplace/trust/web-research - Skill page: https://superagentskill.com/marketplace/web-research - Live version (always current) via MCP: https://superagentskill.com/api/mcp Reinstall or update with `npx skills update`, or pull the live graded version with `npx super-agent install web-research`.
在 GitHub 查看