research-expert

Conducts structured web research with mode-based search depth and file-based report output.

3|Updated Jan 15, 2026
One-click install
npx skills add https://github.com/trudyan141/my-antigravity-agents-kit --skill research-expert-trudyan141
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-expert
Source: https://github.com/trudyan141/my-antigravity-agents-kit/tree/main/templates/.agent/skills/research-expert
Command: npx skills add https://github.com/trudyan141/my-antigravity-agents-kit --skill research-expert-trudyan141

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Gathering reliable information from the web often produces scattered, unsourced results or bloated context windows. This Skill structures research into defined modes with controlled search budgets and writes full findings to files, returning only concise summaries. ## Core Features & Use Cases - Mode-Based Research Depth: Automatically detects Quick Verification, Focused Investigation, or Deep Research mode from task keywords and adjusts tool-call budgets accordingly. - Source Quality Evaluation: Ranks sources from primary documentation to general web content, flags unreliable or outdated material, and documents contradictions. - Filesystem Report Output: Writes comprehensive markdown reports with findings, sources, gaps, and methodology to /tmp files, returning only a lightweight summary to preserve context. - Use Case: Ask the agent to verify a technical claim, investigate a library's capabilities, or produce a deep-dive analysis of a market trend, and receive a sourced, structured report saved to disk. ## Quick Start Research the current state of transformer attention mechanisms and save a comprehensive sourced report to a file.

Frequently Asked Questions about research-expert

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a deep research task with an AI agent?

Include keywords like "comprehensive", "thorough", or "deep dive" in your request. The skill detects Deep Research mode and allocates 10-15 tool calls for exhaustive exploration, producing a detailed multi-section report saved to a file.

How to verify a single fact quickly using web search?

Phrase the request with words like "verify", "confirm", or "quick check". The skill enters Quick Verification mode, using only 3-5 tool calls to find authoritative confirmation and returning a brief sourced answer.

Does the research output include source citations?

Yes, every major claim includes inline attribution with direct quotes, source titles, URLs, and publication dates. The report also documents contradictions between sources and explains which sources are most credible.

Where are the research reports saved?

Full reports are written to /tmp with filenames like research_20240328_topic.md. Only a short summary with the file path, key topics, and source count is returned to the conversation to avoid token overflow.

What happens when web research finds no results?

The skill reports the failure clearly, listing the search queries attempted. It also documents inaccessible sources, conflicting information with all versions cited, and gaps requiring further investigation.