Web research

Skill sumitaich1998/jarvisvr/skills/research/web-research

An AI agentic operating system for mixed reality on the Meta Quest 3 — your own J.A.R.V.I.S. Multi-agent orchestration, multimodal perception (sight/hearing/gaze), 42 holographic widgets, 20 LLM providers.

Install

npx -y skills add sumitaich1998/jarvisvr --skill web-research

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Look something up on the web and present a sourced answer, optionally opening the top result in a spatial browser. Use for factual questions, definitions, how-to lookups, comparisons, and "search for…/look up…/find information on…". Triggers: search, look up, find info, what is, who is, how do I, research, google, current weather/facts.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.7 KB, as published. Nobody here has run it

Web Research

Turn a question into a short, sourced spoken answer plus an optional artifact the user can open and keep.

Steps

Form a tight query. Extract entities and intent; drop filler. For "current X" questions (weather, scores, prices) prefer the dedicated specialist (research-agent's market-briefing, or web-research for general facts).
Call web_search{query}. Returns data.results (title, snippet, url) and a web_panel directive for the top hit.
Synthesize a 1–2 sentence answer in agent.speech; cite the source name.
Surface an artifact: keep the web_panel for browsing, or summarize the findings onto a panel (show_panel) when reading beats browsing.
If the user wants depth on one result, hand to summarize-source.

Output

agent.speech:

{ "text": "Mixed reality blends real and virtual so they coexist and interact —
  here's the overview from Wikipedia.", "final": true }

web_panel (show_web / from web_search):

{ "widget_type": "web_panel",
  "props": { "url": "https://en.wikipedia.org/wiki/Mixed_reality", "title": "Search: mixed reality",
             "interactive": true } }

Summary alternative (show_panel → panel):

{ "widget_type": "panel",
  "props": { "title": "Mixed reality", "body": "Definition, history, and devices.",
             "sections": [ { "heading": "Top sources", "text": "Wikipedia, Nature, BBC" } ] } }

Edge cases

No/low-quality results → say what you tried and offer to refine the query; don't fabricate facts or URLs.
Time-sensitive ("today", "latest") → note recency limits when offline; live fetch requires a configured provider.
Multi-source synthesis (compare 3 articles) → delegate to summarizer sub-agents (one per source) and merge — see orchestration agent-routing.
News specifically → use news-digest; markets → market-briefing.
Reading vs browsing → long answers belong on a panel/document_viewer, not read aloud in full.